Java EE, Web Services, WS-BPEL, SOA

AMQP: RabbitMQ, Spring, Apache Camel, and Apache Qid

2012-02-02T11:07:00.001+01:00

As you know I'm open-sourcing and completely overhauling my PhD system. One of my goals was to replace internal JMS queues with AMQP. Today I'll show you how I did it and why I was forced to change RabbitMQ to Apache Qpid.

AMQP

In short. AMQP is an open standard application layer protocol for message-oriented middleware. The most important feature is that AMQP is a wire-level protocol and is interoperable by design. JMS is just an API. Altough JMS brokers can be used in .NET applications (see my post: ActiveMQ and .NET combined!), but the whole JMS specification does not guarantee interoperability. Also, the AMQP is by design more flexible and powerful (e.g., supports two-way communication by design) - they simply learnt from JMS mistakes :).

Oh, forgot to mention. The AMQP was originally developed by banks :) so I don't have to say that AMQP is secure, fault-tolerant, and so on.

RabbitMQ

RabbitMQ is the most mature AMQP broker. AMQP is written in Erlang so you have to download it first (RabbitMQ Windows installer does it for you). Download it from here: http://www.rabbitmq.com/.

I also recommend installing the web management console. From Rabbit's sbin directory execute:

rabbitmq-plugins enable rabbitmq_management

If you're on Windows and you installed Rabbit service you have to restart it.

That's it.

Spring

Well, it turned out that VMware bought RabbitMQ and SpringSource developers are now developing it. Given this fact, you shouldn't be surprised that Spring - RabbitMQ integration is childishly simple.

Add spring-rabbit dependency to your Maven project, and then in Spring configuration paste the following:

<rabbit:connection-factory id="connectionFactory" />
<rabbit:template connection-factory="connectionFactory" id="amqpTemplate" routing-key="myqueue" />
<rabbit:admin connection-factory="connectionFactory" />
<rabbit:queue name="myqueue" />

The default configuration assumes that RabbitMQ is running on a local server using the default port and default credentials (guest/guest). Of course all these settings are configurable.

To sent a message to "myqueue" queue just inject an instance of AmqpTemplate into your service and send the message. An example would be:

@Service
public class HomeController {
    @Autowired
    private AmqpTemplate amqpTemplate;
    public void sendMessage(Bundle bundle) throws IOException {       
        byte[] body = IOUtils.toByteArray(bundle.getInputStream());
        MessageProperties messageProperties = new MessageProperties();
        messageProperties.setContentType(bundle.getContentType());
        messageProperties.setContentLength(bundle.getSize());
        messageProperties.setTimestamp(new Date());
        messageProperties.setDeliveryMode(MessageDeliveryMode.PERSISTENT);
        Message message = new Message(body, messageProperties);
        amqpTemplate.send(message);
    }
}

You can open the web console http://localhost:55672/mgmt/ and see 1 message in "myqueue" queue.

Apache Camel

To read a message from Apache Camel you first have to add camel-amqp dependency to your POM. Then just copy and paste the following route definition:

<camelContext xmlns="http://camel.apache.org/schema/spring">
 <route>
  <from uri="amqp:queue:myqueue" />
  <to uri="log:Message" />
 </route>
</camelContext>

Run the route by executing mvn:camel-run and... you'll see an error.

Cutting long story short. Apache Camel 2.9.0 doesn't work with RabbitMQ. This is because camel-amqp component is using Apache Qpid client under the hood. Current Qpid version is 0.14, but Qpid guys forgot to upload new jars to Maven public repo. Thus camel-amqp is still using Qpid 0.12 whose client doesn't seem to negotiate protocols. Even if you exclude qpid-commons and qpid-client dependencies and explicitly add Qpid 0.14 ones (download them and install in your local repo) there will be an exception thrown from camel-amqp component as there is no longer a default ConnectionFactory constructor.

Thus I was forced to install Qpid.

Qpid

I downloaded the Java server and simply run it. There is no web management console, but that's OK. You can use JConsole for JMX.

Spring AMQP and Qpid

In order to make Spring AMQP work with Qpid copy and paste the following configuration:

<camelContext xmlns="http://camel.apache.org/schema/spring">
 <route>
  <from uri="amqp:queue:queue" />
  <to uri="log:Message" />
 </route>
</camelContext>

<bean id="amqp" class="org.apache.camel.component.amqp.AMQPComponent">
 <property name="connectionFactory" ref="amqConnectionFactory" />
</bean>

<bean id="amqConnectionFactory" class="org.apache.qpid.client.AMQConnectionFactory">
 <property name="host" value="localhost" />
 <property name="port" value="5672" />
 <property name="defaultUsername" value="guest" />
 <property name="defaultPassword" value="guest" />
 <property name="virtualPath" value="/development" />
</bean>

As you can see in the above snippet I explicitly created AMPQComponent with connectionFactory set to Apache Qpid AMQConnectionFactory object.

Source code and working example

This solution is a part of the Qualitas project. I use Spring MVC to handle uploads of business processes bundles (e.g., zipped archive of a WS-BPEL process) and send it to AMQP queue. Then Apache Camel consumes the message, does additional processing of the bundle, and installs it on a remote business process execution engine.

The projects you are most interested in are:

qualitas-webapp (Spring MVC sending message to AMQP)
qualitas-internall-installation (Apache Camel route consuming messages from AMQP)

To check out 0.0.2-SNAPSHOT tag from here: http://code.google.com/p/qualitas/source/browse/.

Qualitas

Read more about Qualitas project here: http://code.google.com/p/qualitas/. Happy to welcome new developers on board!

cheers,
Łukasz

Resolutions - 2012 Edition

2012-01-23T09:25:00.000+01:00

Everyone has New Year's resolutions. Here is mine: open-source and completely overhaul my PhD system.

The original system

The original system (code name: MS DIES) is written in Java EE 5. The frontent is JSF 1.2 (Facelets, MyFaces, Tomahawk) and the backend is EJB 3.0 (Web Services, SLSB, MDB). The original system runs on Apache Geronimo 2.2 and uses IBM DB2 pureXML as its database.

Here are some screenshots of the original system: http://code.google.com/p/qualitas/wiki/ScreenshotsOfOldSystem.

The new system

New system got a new name: Qualitas. It will be more lightweight (Spring 3.0 and/or Geronimo 3.0 + Aries). I also would like to replace all internal processes (Web Services, EJB, MDB) with Apache Camel routes and use some other sexy technologies (JBoss Drools for computing quality scores, AMPQ instead of JMS, etc.). I already know IBM DB2 pureXML so I'm considering evaluating Oracle and its XML support.

Short term objectives

I'm especially tempted with using Apache Geronimo 3.0 and Aries! But that's a longer term objective. The very first one would be to create a Little-Q distribution which will contain all core processes and will offer means of embedding it into other products. On top of Little-Q I plan to create the complete system, the fully fledged Qualitas. The front-end would most probably look like the orginal one (screenshots above show MyFaces + Tomahawk and maybe I will use RichFaces or some other JSF + AJAX stuff).

I would like to release Little-Q in 6 months time. The prototype will support gathering, analysing, and computing quality scores for WS-BPEL processes running on Apache ODE engine.

Project info

The Qualitas is hosted on Google Code: https://code.google.com/p/qualitas/. Its license is Apache License 2.0. So you can do what ever you like with it! Follow to get updates on the progress. And yes, you are most welcomed to join the project!

cheers,
Łukasz

iOS and Android push notifications from Java

2011-11-25T09:55:00.001+01:00

For a couple of past weeks I have been working as a Technical Architect in a mobile application for a large bank (or should I say an insurance department of that bank). The application was written in PhoneGap and Sencha Touch. Two platforms were supported: iOS and Android. I was responsible (apart of many other things) for developing notifications for iOS and Android. The notification simply told the insurance agent how many cases & actions were updated/changed.

Sending notifications to iOS devices

In order to use Apple Push Notification Service (APNS), apart of standard developer's certificate and a valid provisioning profile, you need one more thing. You need the APNS ceritificate which is used for mutual authorisation. You can generate it from Apple's iOS Provisioning Portal. Once you have it, you're ready to send notifcations.

APNS is using a binary protocol. Of course you don't have to implement it from the scratch. You can use a very good and throughtly tested notnoop java-apns library which is available here: https://github.com/notnoop/java-apns.

Using it is extremelly simple, the following code works fine (Notification is my internal class):

@Service
@Qualifier("Apns")
public class ApnsPushNotificationServiceImpl implements PushNotificationService {
 private Resource certResource;
 private String certPassword;
 private boolean useProductionServer;
 private String serverAddress;
 private static final int DEFAULT_APNS_PORT = 2195;
 public void setCertResource(Resource certResource) {
  this.certResource = certResource;
 }
 public void setCertPassword(String certPassword) {
  this.certPassword = certPassword;
 }
 public void setUseProductionServer(boolean useProductionServer) {
  this.useProductionServer = useProductionServer;
 }
 public void setServerAddress(String serverAddress) {
  this.serverAddress = serverAddress;
 }
 public void pushNotification(Notification notification) {
  List<Notification> notList = new ArrayList<Notification>();
  notList.add(notification);
  pushNotifications(notList);
 }
 public void pushNotifications(List<Notification> notifications) {
  ApnsService service = createApnsService();
  service.start();
  for (Notification notification : notifications) {
   doSendNotification(notification, service);
  }
  service.stop();
 }
 private ApnsService createApnsService() {
  ApnsService service = null;
  ApnsServiceBuilder serviceBuilder = APNS.newService()
    .withCert(certResource.getInputStream(), certPassword);
  if (null != serverAddress && serverAddress.length() > 0) {
   serviceBuilder.withGatewayDestination(serverAddress,
     DEFAULT_APNS_PORT);
  } else if (useProductionServer) {
   serviceBuilder.withProductionDestination();
  } else {
   serviceBuilder.withSandboxDestination();
  }
  service = serviceBuilder.build();
  return service;
 }
 private void doSendNotification(Notification notification,
   ApnsService service) {
  PayloadBuilder payloadBuilder = APNS.newPayload();
  payloadBuilder = payloadBuilder.badge(notification.getBadge());
  payloadBuilder = payloadBuilder.sound(notification.getSound());
  if (notification.getBody() != null) {
   payloadBuilder = payloadBuilder.alertBody(notification.getBody());
  }
  String payload = payloadBuilder.build();
  service.push(notification.getDeviceToken(), payload);
 }

Sending notification to Android devices

In comparison to iOS notifications, implemeting the Android Cloud 2 Device Messaging (AC2DM) was a piece of cake. I didn't need any third part library. Thanks to simple and clear Google's RESTful services, I was able to write a code responsible for pushing notifications to Android devices in less than an hour.

The whole process comprises of 2 steps. The first one is authentication, the second one is the actual pushing of a notification. The source code is as follows:

@Service
@Qualifier("Ac2dm")
public class Ac2dmPushNotificationServiceImpl implements PushNotificationService {
 private String sendingRoleAccount;
 private String sendingRolePassword;
 public void pushNotification(Notification notification) {
  List<Notification> notList = new ArrayList<Notification<();
  notList.add(notification);
  pushNotifications(notList);
 }
 public void pushNotifications(List<Notification> notifications) {
  try {
   HttpClient client = new HttpClient();
   PostMethod method = new PostMethod(
     "https://www.google.com/accounts/ClientLogin");
   method.addParameter("Email", sendingRoleAccount);
   method.addParameter("Passwd", sendingRolePassword);
   method.addParameter("accountType", "HOSTED_OR_GOOGLE");
   method.addParameter("source", "unit-test");
   method.addParameter("service", "ac2dm");
   client.executeMethod(method);
   byte[] responseBody = method.getResponseBody();
   String response = new String(responseBody);
   String auth = response.split("\n")[2];
   String token = auth.split("=")[1];
   for (Notification notification : notifications) {
    doSendNotification(notification, token, client);
   }
  } catch (Throwable t) {
   throw new RuntimeException(t);
  }
 }
 private void doSendNotification(Notification notification, String token, HttpClient client) throws HttpException, IOException {
  PostMethod method = new PostMethod("https://android.apis.google.com/c2dm/send");
  method.addParameter("registration_id", notification.getDeviceToken());
  method.addParameter("collapse_key", "collapse");
  method.addParameter("data.payload", notification.getBadge().toString());  
  Header header = new Header("Authorization", "GoogleLogin auth="+token);
  method.addRequestHeader(header);
  client.executeMethod(method);  
  byte[] responseBody = method.getResponseBody();  
  String response = new String(responseBody);
 }
 public void setSendingRoleAccount(String sendingRoleAccount) {
  this.sendingRoleAccount = sendingRoleAccount;
 }
 public void setSendingRolePassword(String sendingRolePassword) {
  this.sendingRolePassword = sendingRolePassword;
 }

A Note on WebSphere & IBM's Java 5

When the application was deployed to the UAT server and we tried to run integration test responsible for pushing notification to one of our test iPhones, we saw "java.security.InvalidKeyException: Illegal key size" message in the log. After many hours of investigation it turned out that, IBM's Java 5 supports only 512 key sizes. We had to update IBM's Java 5 security jars in order to make it read the APNS certificate.

-1 for IBM!

Summary

It took me some time, but when you see a notifications coming to your device, then using PhoneGap, you launch a JavaScript application which looks like a native app, it makes you feel good :)

cheers,
Łukasz

Customising Spring Security (for mobile application)

2011-10-11T10:37:00.000+02:00

Last time I used Spring Security it was called Acegi Security. So quite a long time ago.

I'm currently TA in 2 projects and one of them is a mobile application written in Phone Gap. This application communicates with a backend server using RESTful Web Services.

One of my task was to write a security module for our application.

UserDetailsService

My first task was to write UserDetailsService. We didn't have and didn't want to have Spring-specific user and role tables. In many Java EE application servers you have to write custom SQL which returns you users and roles. Spring went 1 step farther. You don't have to write any SQL selects, you can implement UserDetailsService interface and inject your DAOs there. For me this a great and powerful feature! So, all I had to do was to implement this 1 method:

UserDetails loadUserByUsername(String username) throws UsernameNotFoundException, DataAccessException;

It took me a few seconds to implement. I injected @Repository DAO objects and simply called the find methods. Then I created an instance of Spring-specific UserDetails object and simply returned it.

Next thing was to define authentication manager in Spring context file with a reference to my mobileUserDetailsService:

<security:authentication-manager>
 <security:authentication-provider user-service-ref="mobileUserDetailsService">
  <security:password-encoder hash="sha-256" />
 </security:authentication-provider>
</security:authentication-manager>

And I was done.

Method-level security

The second thing I wanted to do was to secure method calls instead of URL patterns (as you know RESTful URLs have common parts). I decided to use Spring's @Secured("ROLE_XXX") annotations. To enable method-level security all I had to do was to add the following XML element to Spring configuration file:

<security:global-method-security secured-annotations="enabled">
</security:global-method-security>

Nice.

Exposing authentication API over RESTful Web Services

The very last thing was to use a expose Spring's authentication API over RESTful Web Services. I wrote a MobileAuthenticationManager which was called by my RESTful Web Service.

I only had one problem with Spring not being able to inject AuthenticationManager becuase there were 2 instances of it in my project. I decided to implement ApplicationContextAware interface and inside the setApplicationContext() method retrieve manually org.springframework.security.authenticationManager bean. It looked like this:

@Service
public class MobileAuthenticationManagerImpl implements
  MobileAuthenticationManager, ApplicationContextAware {
 protected AuthenticationManager authenticationManager;
 @Autowired
 protected MobDeviceDao mobDeviceDao;
 public Authentication authenticate(String username, String pin) {
  Authentication request = new UsernamePasswordAuthenticationToken(username, pin);
  Authentication result = null;
  try {
   result = authenticationManager.authenticate(request);
  } catch (BadCredentialsException e) {
   // some more code here
  }
  SecurityContextHolder.getContext().setAuthentication(result);
  // some more code removed here
  return result;
 }
 public boolean clearSecurityContext(String deviceId, String pin) {  
  invalidateSecurityContext();  
  return true;
 } 
 public void invalidateSecurityContext() {
  SecurityContextHolder.getContext().setAuthentication(null);
 }
 public void setApplicationContext(ApplicationContext applicationContext)
   throws BeansException {
  authenticationManager = (AuthenticationManager) applicationContext.getBean("org.springframework.security.authenticationManager");
 }
}

Login and invalid session pages

The very last thing was to change default login-page and invalid-session-url to /logout.json file which contained a JSON logout response.

<security:http create-session="always" >
 <security:form-login login-page="/logout.json"  />
 <security:session-management invalid-session-url="/logout.json" />
</security:http>

Summary

I have to say that Spring (Acegi) Security changed a lot during those nearly 3.5 years when I last used it! It was a piece of cake to write a fully custom security module. Nice work Spring developers!

cheers,
Łukasz

Spring.NET superiority over MS Unity and Ninject - proven

2011-08-26T14:03:00.002+02:00

I'm evaluating .NET-related technologies for the purpose of my latest project (desktop application for processing radiological and pathological images). Today I played with MS Unity and Ninject.

I was quite surprised that either MS Unity nor Ninject are unable to handle bi-directional references, like these:

class A
{
   B B { get; set; }
}

class B 
{
   A A { get; set; }
}

Both MS Unity and Ninject enter infinite loop (create default constructor and set a break point there).

Spring.NET resolves dependencies without any problems. A huge plus for Spring.NET.

cheers,
Łukasz

p.s.
In case someone would argue why I use bi-directional references: it's a desktop application where implementing MVC/MVP requires view to have a reference to controller/presenter and controller/presenter to have a reference to view.

p.s.2
I used latest versions of Unity and Ninject available in NuGet.

Aspect Oriented Programming in .NET using PostSharp

2011-08-24T11:00:00.001+02:00

PostSharp is claiming to be the most comprehensive AOP library in the .NET world. I gave it a try and I must to say that their claim is very well founded.

Read more to see PostSharp in action.

PostSharp

PostSharp is not a free tool, but the trail version should be OK to get you started and do some proof of concepts.

The best way to learn AOP and PostSharp itself is to watch PostSharp's podcasts. The usecases shown in those podcasts are well balanced, they are not too easy not too difficult. I suspect that even people without any previous AOP experience will get a pretty solid idea of what AOP is. The podcasts are here: http://www.sharpcrafters.com/media/tutorials.

Audit aspect

I wanted to write a fine-grained audit aspect which could log execution traces. I wanted to be able to do something like this.

1) Add [Audit] attribute to methods:

public class GenericDAO<T>
{
    [Audit("DAO")]
    public void Save(T entity)
    {
        /* ... */
    }
}

2) Add [Audit] attribute to classes, but be able to switch off auditing for some of the methods using [NoAudit] attribute just like this:

[Audit("DAO")]
public class GenericDAO<T>
{
    [NoAudit]
    public GenericDAO()
    {
        /* ... */
    }
}

It took me 10 minutes to write the following 2 aspects (note if I were to use them in production, which I'm sure I will in a few weeks, I would make them a slightly more optimised - like reduce reflections etc). Here they are:

[Serializable]
public sealed class AuditAttribute : OnMethodBoundaryAspect
{
    private readonly string _category;
    public AuditAttribute(string category)
    {
        _category = category;
    }
    public override bool CompileTimeValidate(MethodBase method)
    {
        object[] attributes = method.GetCustomAttributes(typeof(NoAuditAttribute), true);
        if (attributes.Length > 0)
        {
            Message.Write(SeverityType.Warning, "Audit", "Skipping auditing on " + method.DeclaringType.ToString() + " method " + method.ToString());
            return false;
        }
        return base.CompileTimeValidate(method);
    }
    public override void OnEntry(MethodExecutionArgs args)
    {
        StringBuilder sb = new StringBuilder();
        ParameterInfo[] parameterInfos = args.Method.GetParameters();
        for (int i = 0; i < parameterInfos.Length; i++)
        {
            ParameterInfo parameterInfo = parameterInfos[i];
            object parameterValue = args.Arguments[i] ?? "null";
            sb.AppendFormat("{0}={1}", parameterInfo.Name, parameterValue);
            if (i < parameterInfos.Length - 1)
            {
                sb.Append(", ");
            }
        }
        Trace.WriteLine(string.Format("Entering {0}.{1} {2}",
            args.Method.DeclaringType.Name, args.Method.Name, sb), _category);
        args.MethodExecutionTag = DateTime.Now;
    }
    public override void OnException(MethodExecutionArgs args)
    {
        Trace.WriteLine(string.Format("Exception " + args.Exception + " thrown in {0}.{1}",
            args.Method.DeclaringType.Name, args.Method.Name), _category);
    }
    public override void OnSuccess(MethodExecutionArgs args)
    {
        DateTime endTime = DateTime.Now;
        TimeSpan executionTime = endTime.Subtract((DateTime)args.MethodExecutionTag);
        bool isVoid = ((MethodInfo)args.Method).ReturnType == typeof(void);
        object returnValue;
        if (isVoid)
        {
            returnValue = "void";
        }
        else
        {
            returnValue = args.ReturnValue ?? "null";    
        }
        Trace.WriteLine(string.Format("Leaving {0}.{1} processing took me {2}ms return value was {3}",
            args.Method.DeclaringType.Name, args.Method.Name, executionTime, returnValue), _category);
    }
}
[Serializable]
public sealed class NoAuditAttribute : Attribute
{
}

When AuditAttribute is applied to class and a method has NoAuditAttribute then a warning message "Skipping auditing on ClassABC method MethodXYZ" is displayed in MS VS errors tab (see CompileTimeValidate method above).

Applying different aspects to the same method

Next I wanted to mix my aspects. Apart of auditing aspect I wrote a ValidateAttribute and wanted to use it this way:

[Serializable]
public sealed class ValidateAttribute : OnMethodBoundaryAspect
{
    /* ... */
}
[Audit("DAO")]
public class GenericDAO<T>
{
    [Validate("Full")]
    public void Save(T entity)
    {
        /* ... */
    }
}

There was one small issue with the above code. It worked, but PostSharp was saying that the order of aspects may have been undeterministic. But this is really not a problem with PostSharp. If I wanted to log the parameters before validation we should have added AspectTypeDependency attribute to ValidateAttribute just like this:

[AspectTypeDependency(AspectDependencyAction.Order, AspectDependencyPosition.After, typeof(AuditAttribute))]

Summary

As an excercise you can write support for shadowing parameters. For example add a new parameter to the AuditAttribute which would take an array of parameters' names to shadow (like username, password) and print either null if passed value is null or "*****" otherwise.

cheers,
Łukasz

MS Enterprise Library Logging

2011-08-18T15:50:00.000+02:00

Yesterday I played a little bit with MS Enterprise Library Validation. Today I tried the Logging Application Block. I like it. The number of options is just overwhelming. You can log to files, rolling files, system diagnostics, database, SMTP, MS queues, WMI, WCF, and I probably missed a few more.

Today's post is a simple how-to post.

Project and database setup

For the database trace listener I had to create 3 tables and 4 stored procedures. The SQL statements are stored in C:\path\to\your\EntLib50Src\Blocks\Logging\Src\DatabaseTraceListener\Scripts\LoggingDatabase.sql file. Apart of creating a database I had to create a "Database block" and define a valid connection string. And that was all.

Logging configuration

I opened the visual editor and added Logging block. It took me 5 minutes to configure the following trace listeners: rolling file, SMTP, and database.

My configuration file looked like this:

<?xml version="1.0" encoding="utf-8" ?>
<configuration>
    <configSections>
        <section name="loggingConfiguration" type="Microsoft.Practices.EnterpriseLibrary.Logging.Configuration.LoggingSettings, Microsoft.Practices.EnterpriseLibrary.Logging" requirePermission="true" />
        <section name="dataConfiguration" type="Microsoft.Practices.EnterpriseLibrary.Data.Configuration.DatabaseSettings, Microsoft.Practices.EnterpriseLibrary.Data" requirePermission="true" />
    </configSections>
    <loggingConfiguration name="" tracingEnabled="true" defaultCategory="General">
        <listeners>
            <add name="Rolling Flat File Trace Listener" type="Microsoft.Practices.EnterpriseLibrary.Logging.TraceListeners.RollingFlatFileTraceListener, Microsoft.Practices.EnterpriseLibrary.Logging"
                listenerDataType="Microsoft.Practices.EnterpriseLibrary.Logging.Configuration.RollingFlatFileTraceListenerData, Microsoft.Practices.EnterpriseLibrary.Logging"
                formatter="Text Formatter" traceOutputOptions="LogicalOperationStack, DateTime, Timestamp, ProcessId, ThreadId, Callstack" />
            <add name="Error Rolling Flat File Trace Listener" type="Microsoft.Practices.EnterpriseLibrary.Logging.TraceListeners.RollingFlatFileTraceListener, Microsoft.Practices.EnterpriseLibrary.Logging"
                listenerDataType="Microsoft.Practices.EnterpriseLibrary.Logging.Configuration.RollingFlatFileTraceListenerData, Microsoft.Practices.EnterpriseLibrary.Logging"
                fileName="error-rolling.log" formatter="Text Formatter" traceOutputOptions="LogicalOperationStack, DateTime, Timestamp, ProcessId, ThreadId, Callstack" />
            <add name="Email Trace Listener" type="Microsoft.Practices.EnterpriseLibrary.Logging.TraceListeners.EmailTraceListener, Microsoft.Practices.EnterpriseLibrary.Logging"
                listenerDataType="Microsoft.Practices.EnterpriseLibrary.Logging.Configuration.EmailTraceListenerData, Microsoft.Practices.EnterpriseLibrary.Logging"
                toAddress="lukasz@xxx" fromAddress="from@example.com"
                smtpServer="exchange.xxx" formatter="Text Formatter"
                traceOutputOptions="LogicalOperationStack, DateTime, Timestamp, ProcessId, ThreadId, Callstack" />
            <add name="Database Trace Listener" type="Microsoft.Practices.EnterpriseLibrary.Logging.Database.FormattedDatabaseTraceListener, Microsoft.Practices.EnterpriseLibrary.Logging.Database"
                listenerDataType="Microsoft.Practices.EnterpriseLibrary.Logging.Database.Configuration.FormattedDatabaseTraceListenerData, Microsoft.Practices.EnterpriseLibrary.Logging.Database"
                databaseInstanceName="SqlServer" writeLogStoredProcName="WriteLog"
                addCategoryStoredProcName="AddCategory" formatter="" traceOutputOptions="LogicalOperationStack, DateTime, Timestamp, ProcessId, ThreadId, Callstack" />
        </listeners>
        <formatters>
            <add type="Microsoft.Practices.EnterpriseLibrary.Logging.Formatters.TextFormatter, Microsoft.Practices.EnterpriseLibrary.Logging"
                template="Timestamp: {timestamp}{newline}
Message: {message}{newline}
Category: {category}{newline}
Priority: {priority}{newline}
EventId: {eventid}{newline}
Severity: {severity}{newline}
Title:{title}{newline}
Machine: {localMachine}{newline}
App Domain: {localAppDomain}{newline}
ProcessId: {localProcessId}{newline}
Process Name: {localProcessName}{newline}
Thread Name: {threadName}{newline}
Win32 ThreadId:{win32ThreadId}{newline}
Extended Properties: {dictionary({key} - {value}{newline})}"
                name="Text Formatter" />
        </formatters>
        <logFilters>
            <add type="Microsoft.Practices.EnterpriseLibrary.Logging.Filters.CategoryFilter, Microsoft.Practices.EnterpriseLibrary.Logging"
                categoryFilterMode="DenyAllExceptAllowed" name="Category Filter">
                <categoryFilters>
                    <add name="General" />
                </categoryFilters>
            </add>
        </logFilters>
        <categorySources>
            <add switchValue="All" name="General">
                <listeners>
                    <add name="Rolling Flat File Trace Listener" />
                </listeners>
            </add>
        </categorySources>
        <specialSources>
            <allEvents switchValue="All" name="All Events">
                <listeners>
                    <add name="Rolling Flat File Trace Listener" />
                    <add name="Database Trace Listener" />
                </listeners>
            </allEvents>
            <notProcessed switchValue="All" name="Unprocessed Category" />
            <errors switchValue="All" name="Logging Errors & Warnings">
                <listeners>
                    <add name="Error Rolling Flat File Trace Listener" />
                </listeners>
            </errors>
        </specialSources>
    </loggingConfiguration>
    <dataConfiguration defaultDatabase="SqlServer" />
    <connectionStrings>
        <add name="SqlServer" connectionString="data source=.\SQLEXPRESS1;Database=Logging;Trusted_Connection=True;MultipleActiveResultSets=True"
            providerName="System.Data.SqlClient" />
    </connectionStrings>
</configuration>

Source code

Then I just used the LogWriter in my code:

var source = new FileConfigurationSource("ModuleConfiguration.config");

var logWriter = new LogWriterFactory(source).Create();

logWriter.Write("Message 1", "BBB Category");
logWriter.Write("Message 2", "General");
logWriter.Write("Message 3", "AAA Category");

Summary

Simple isn't it? The above code uses an external configuration file. When you use the logging configuration embedded in your App.config it's even simpler as you don't have to use source objects and log factories - you just log entries using static method Logger.Write().

Next thing. Aspect oriented programming and PostSharp!

thanks,
Łukasz

MS Enterprise Library Validation

2011-08-17T12:25:00.001+02:00

Last week I played around with Spring.NET Validation. Today I gave MS Enterprise Library a try. In my opinion it's much more powerful than Spring.NET's counterpart. Read below to see why.

Getting Enterprise Library

Simply download and install Enterprise Library 5.0. Note: to write code which uses Enterprise Library you can use NuGet packages, but then you won't get things like a visual XML configuration editor (integrated with MS VS of course).

XML configuration approach

Using the visual editor I was able to create validation rules for class Person. Apart of simple validation rules I was able to create rule sets. Finally, I ended up with defining a "Full" rule set that requires all fields not to be null and EmailAddress property to be a valid email and a "Partial" rule set which allowed EmailAddress to be null.

The configuration looked like this:

<validation>
    <type name="BusinessModuleTest.Person" defaultRuleset="Full"
        assemblyName="BusinessModuleTest, Version=1.0.0.0, Culture=neutral, PublicKeyToken=null">
        <ruleset name="Full">
            <properties>
                <property name="FirstName">
                    <validator type="Microsoft.Practices.EnterpriseLibrary.Validation.Validators.NotNullValidator, Microsoft.Practices.EnterpriseLibrary.Validation"
                        messageTemplate="First name cannot be null" name="FirstName Not Null Validator" />
                </property>
                <property name="EmailAddress">
                    <validator type="Microsoft.Practices.EnterpriseLibrary.Validation.Validators.RegexValidator, Microsoft.Practices.EnterpriseLibrary.Validation"
                        pattern="\w+([-+.']\w+)*@\w+([-.]\w+)*\.\w+([-.]\w+)*" options="IgnoreCase"
                        messageTemplate="EmailAddress is not valid" name="Regular Expression Validator" />
                </property>
            </properties>
        </ruleset>
        <ruleset name="Partial">
            <add type="Microsoft.Practices.EnterpriseLibrary.Validation.Validators.NotNullValidator, Microsoft.Practices.EnterpriseLibrary.Validation"
                messageTemplate="First name cannot be null" name="FirstName Not Null Validator" />
        </ruleset>
    </type>
</validation>

Then I wrote the following NUnit test:

[TestFixture]
public class EntLibValidationTest
{
    [Test]
    public void TestValidateTrue()
    {
        var source = new FileConfigurationSource("ModuleConfiguration.config");
        var validator = ValidationFactory.CreateValidatorFromConfiguration<Person>(source);
        var person = new Person() { FirstName = "Łukasz", EmailAddress = "lukasz@server.pl" };
        ValidationResults r = validator.Validate(person);
        Assert.IsTrue(r.IsValid);
    }
    [Test]
    public void TestValidateFalse()
    {
        var source = new FileConfigurationSource("ModuleConfiguration.config");
        var validator = ValidationFactory.CreateValidatorFromConfiguration(typeof(Person), "Full", source);
        var person = new Person() { FirstName = "Łukasz" };
        ValidationResults results = validator.Validate(person);
        Assert.IsFalse(results.IsValid);
        Assert.IsTrue(results.Count == 1);
        foreach (ValidationResult result in results)
        {
            Assert.AreEqual("EmailAddress", result.Key);
        }
    }
    [Test]
    public void TestValidatePartial()
    {
        var source = new FileConfigurationSource("ModuleConfiguration.config");
        var validator = ValidationFactory.CreateValidatorFromConfiguration(typeof(Person), "Partial", source);
        var person = new Person() { FirstName = "Łukasz" };
        ValidationResults results = validator.Validate(person);
        Assert.IsTrue(results.IsValid);
    }
}
public class Person
{
    public string FirstName { get; set; }
    public string EmailAddress { get; set; }
}

All worked as expected.

Attributes approach

Next I wanted to try the attributes approach. I wrote Person2 class which looked like this:

public class Person2
{
    [NotNullValidator(ErrorMessage = "First name must not be null", Ruleset = "Full")]
    [NotNullValidator(ErrorMessage = "First name must not be null", Ruleset = "Partial")]
    public string FirstName { get; set; }
    [RegexValidator(@"\w+([-+.']\w+)*@\w+([-.]\w+)*\.\w+([-.]\w+)*", MessageTemplate = "Email address is not valid", Ruleset = "Full")]
    public string EmailAddress { get; set; }
}

A 1:1 mapping of what I defined earlier in XML file.

Next thing was to write the validation test:

[TestFixture]
public class EntLibValidationAttributeTest
{
    [Test]
    public void TestValidateTrue()
    {
        var validator = ValidationFactory.CreateValidatorFromAttributes<Person2>("Full");
        var person = new Person2() { FirstName = "Łukasz", EmailAddress = "lukasz.budnik@gmail.com" };
        ValidationResults r = validator.Validate(person);
        Assert.IsTrue(r.IsValid);
    }
    [Test]
    public void TestValidateFalse()
    {
        var validator = ValidationFactory.CreateValidatorFromAttributes<Person2>("Full");
        var person = new Person2() { FirstName = "Łukasz" };
        ValidationResults results = validator.Validate(person);
        Assert.IsFalse(results.IsValid);
        Assert.IsTrue(results.Count == 1);
        foreach (ValidationResult result in results)
        {
            Assert.AreEqual("EmailAddress", result.Key);
        }
    }
    [Test]
    public void TestValidatePartial()
    {
        var validator = ValidationFactory.CreateValidatorFromAttributes<Person2>("Partial");
        var person = new Person2() { FirstName = "Łukasz" };
        ValidationResults results = validator.Validate(person);
        Assert.IsTrue(results.IsValid);
    }
}

Again, all worked like a charm.

Summary

I must say I like Enterprise Library Validation much more than Spring.NET Validation (which is a surprise for me). Enterprise Library is internationalisation-ready and supports resource messages out of the box (I didn't use them in my simple examples, but I surely will use them in my production application).

My next post will be about Enterprise Library Logging. Then I will move to more interesting parts like AOP with PostSharp and mocking with Moq.

stay tuned,
Łukasz

Spring.NET Validation

2011-08-12T11:31:00.000+02:00

As a part of investigating .NET world I started playing around with Spring.NET. Today I will show you some basics of Spring.NET validation.

Spring.NET Validation Basics

Validation using Spring.NET is pretty straight forward. I created a User class which looked like this:

public class User
{
    public string Name { get; set; }
    public string Email { get; set; }
}

Then I created a NUnit test and wrote first test method. I started with a simple RequiredValidator. As a test parameter I passed "Name" as I wanted to validate Name property. The second parameter is null, I used it later on.

[TestFixture]
public class ValidatorTest
{
    [Test]
    public void TestRequiredValidator()
    {
        User user = new User() { Name = "Łukasz"};
        BaseValidator baseValidator = new RequiredValidator("Name", null);
        IValidationErrors errors = new ValidationErrors();
        bool valid = baseValidator.Validate(user, errors);
        Assert.IsTrue(valid);
    }
}

Test passed.

Maybe something a little bit more complex? Regular expression. Mind my first name is Łukasz and the letter Ł has to be explicitly added to the pattern (otherwise the test fails):

[Test]
public void TestRegexpValidator()
{
    User user = new User() { Name = "Łukasz" };
    BaseValidator baseValidator = new RegularExpressionValidator("Name", null, "[A-Za-zŁł]{2,}");
    IValidationErrors errors = new ValidationErrors();
    bool valid = baseValidator.Validate(user, errors);
    Assert.IsTrue(valid);
}

OK. Nice.

Validator groups

So maybe we could validate more than one property in a single step? Why not. Spring.NET comes with validator groups. There are 3 different validator groups, here I use simple validator group (more in Spring.NET documentation):

[Test]
public void TestValidatorGroupTrue()
{
    User user = new User() { Name = "Łukasz", Email = "lukasz.budnik@gmail.com"};
    ValidatorGroup validatorGroup = new ValidatorGroup();
    validatorGroup.Validators.Add(new RegularExpressionValidator("Name", null, "[A-Za-zŁł]{2,}"));
    validatorGroup.Validators.Add(new EmailValidator("Email", null));
    IValidationErrors errors = new ValidationErrors();
    bool valid = validatorGroup.Validate(user, errors);
    Assert.IsTrue(valid);
}

when parameter

That works fine. But will regular expression validator work for the following scenario: name must not be null and must match the given regular expression?

[Test]
[ExpectedException(typeof(ArgumentException))]
public void TestValidatorGroupException()
{
    User user = new User() { Name = "Łukasz", Email = "lukasz.budnik@gmail.com" };
    ValidatorGroup validatorGroup = new ValidatorGroup();
    validatorGroup.Validators.Add(new RegularExpressionValidator("Name", null, "[A-Za-zŁł]{2,}"));
    validatorGroup.Validators.Add(new EmailValidator("Email", null));
    IValidationErrors errors = new ValidationErrors();
    bool valid = validatorGroup.Validate(user, errors);
    Assert.Fail("There should be an exception thrown by Validate() method");
}

Actually no. The above test will throw an exception (see ExpectedException attribute). In the above example when Name property is null the regular expression validator throws ArgumentException exception. So now what?

Now is the time for the second parameter - the when parameter.

when defines an expression which tells the validator when the validation should be conducted. If the when expression evaluates to false, no action is taken.

In my case when I wanted to check if name was not null and matched the regular expression I had to add two validators and regular expression validator had the when parameter set to "Name != null":

[Test]
public void TestEnhancedValidatorGroup()
{
    User user = new User() { Name = null, Email = null };
    ValidatorGroup validatorGroup = new ValidatorGroup();
    validatorGroup.Validators.Add(new RequiredValidator("Name", null));
    validatorGroup.Validators.Add(new RegularExpressionValidator("Name", "Name != null", "[A-Za-zLl]{2,}"));
    IValidationErrors errors = new ValidationErrors();
    bool valid = validatorGroup.Validate(user, errors);
    Assert.IsFalse(valid);
    user.Name = "L";
    valid = validatorGroup.Validate(user, errors);
    Assert.IsFalse(valid);
    user.Name = "Lukasz";
    valid = validatorGroup.Validate(user, errors);
    Assert.IsTrue(valid);
}

Summary

Quite nice. I just miss a handy profile-based validation, just like it is supported by JSR 303 and frameworks like Oval (I wrote a few posts about Java validation see the validation label).

My next post will be about MS Enterprise Library Validation.

stay tuned,
Łukasz

Spring.NET and NUnit

2011-08-11T12:46:00.000+02:00

As you already know I'm evaluating Spring.NET for a medical desktop software. I did a simple DI, AOP, internationalisation, and logging last week. I wrote about it here: First steps with Spring.NET.

Today I'll focus on testing.

Business logic module

I wrote IGreeterService and IWordDictionaryService interfaces. Then I implemented both interfaces (I don't list dictionary service for brevity):

public class GreeterService : IGreeterService
{
    public IWordDictionaryService Dictionary { get; set; }
    public string Greet(string name)
    {
    	// yes I know, very intelligent use of dictionary service - anyway it's just a test
        bool result = Dictionary.CheckWord("hello");
        return "Hello " + name + "! How are you?";
    }
}

As you can see greeter service has a reference to dictionary service. I will use Spring.NET to inject it.

Spring.NET configuration file

When you are creating a class library project there is no App.config file. Instead you have to create a Context.xml file (or whatever name you want). The root element of this file must be <objects /> element. Also, make sure the Context.xml is an embedded resource. In my case I wrote something like this:

<?xml version="1.0" encoding="utf-8" ?>
<objects xmlns="http://www.springframework.net">
    <object id="GreeterService" type="BusinessModule.Greeter.GreeterService" autowire="byType" />
    <object id="WordDictionaryService" type="BusinessModule.Dictionary.WordDictionaryService" />
    <object type="Spring.Aop.Framework.AutoProxy.ObjectNameAutoProxyCreator, Spring.Aop">
        <property name="ObjectNames">
            <list>
                <value>*Service</value>
            </list>
        </property>
        <property name="InterceptorNames">
            <list>
                <value>SpringLoggingAroundAdvice</value>
            </list>
        </property>
    </object>
    <object name="SpringLoggingAroundAdvice" type="Spring.Aspects.Logging.SimpleLoggingAdvice, Spring.Aop">
        <property name="LogUniqueIdentifier" value="false"/>
        <property name="LogExecutionTime"    value="true"/>
        <property name="LogMethodArguments"  value="true"/>
        <property name="LogReturnValue"      value="true"/>
        <property name="Separator"           value=";"/>
        <property name="LogLevel"            value="Info"/>
        <property name="HideProxyTypeNames"  value="true"/>
        <property name="UseDynamicLogger"    value="true"/>
    </object>
</objects>

Adding test project

Once I had the class library ready I wanted to test it. I added test project to my solution. Apart of adding NUnit DLL references, I added Spring.Core, Spring.Aop, and Spring.Testing.NUnit DLL references. Also, I added my business logic module as a project reference.

I created a simple NUnit test as a subclass of Spring.NET's AbstractDependencyInjectionSpringContextTests class (ConfigLocations property must be overriden).

My test looked like this:

[TestFixture]
public class DictionaryServiceTest : AbstractDependencyInjectionSpringContextTests
{
public IWordDictionaryService DictionaryService { get; set; }
[Test]
public void TestDictionaryServiceTrue()
{
    bool result = DictionaryService.CheckWord("hello");
    Assert.AreEqual(true, result);
}
[Test]
public void TestDictionaryServiceFalse()
{
    bool result = DictionaryService.CheckWord("qwq");
    Assert.AreEqual(false, result);
}
protected override string[] ConfigLocations
{
    get
    {
        return new string[] { "assembly://BusinessModule/BusinessModule.Config/BusinessModuleContext.xml" };
    }
}
}

As you can see, nowhere in my code I refer to ContextRegistry. By default Spring.NET tries to inject all matching objects into test's public properties. So DictionaryService property of type IWordDictionaryService is injected automatically. In a matter of fact you can also inject objects into private fields/properties by setting the PopulateProtectedVariables property to true in test's constructor.

After testing dictionary service I tested greeter service. Everything worked fine. IDictionaryService was autowired by type.

Summary

That would be all. As I suspected testing Spring.NET applications is extremely easy.

My next step would be to use Spring.NET Validation and compare it with MS Enterprise Library Validation.

cheers,
Łukasz

Testing .NET applications with Gallio

2011-08-08T10:37:00.000+02:00

So far I was using MSTest to write and run/debug unit test for my .NET applications. Why? Because MSTest works out of the box in MS VS.

Last week I decided to try out NUnit. The meain reason was a problem with running MSTest tests on a build server. When you want to run MSTests on your build server, the build server complains about missing Microsoft.VisualStudio namespace. In order to run my MSTests on a build server I had to install MS VS! Which is obviously pretty sick.

The second reason was that once I started learning Spring.NET I decided that I would start writing tests in NUnit (Spring.NET supports both MSTest and NUnit).

The third one was simply to try out something which sounded Java-like :)

But this post is not about NUnit but Gallio.

Gallio

NUnit is not supported by MS VS 2010. I started googling in order to find somekind of a plugin for VS. On stackoverflow I found a link with many likes that led me to Gallio.

Gallio turned out to be test automation platform which supported many .NET testing tools. Gallio is not a test framework it's a test runner (and debugger with a little help of MS VS).

I must say that it integrates surprisingly good with MS VS (or should I say ReSharper's Unit Test Runner). You can run/debug test directly from VS. Apart of running tests from VS, Gallio can run tests from command line (Gallio Echo) and Windows GUI (Gallio Icarus which can attach to VS debugger - works like a charm).

Nice. I like it. For those of you who would like to see it in action a short screenshots-loaded official tutorial is here: http://www.gallio.org/wiki/doku.php?id=getting_started:my_first_tests.

Note. I also asked more experienced .NET guys from my company about .NET test platforms. Some of them were using TestDriven.NET. It's not a free tool, but I heard it integrates with MS VS quite nicely.

cheers,
Łukasz

First steps with Spring.NET

2011-08-04T08:51:00.000+02:00

I was appointed as Technical Architect for a C++ desktop (medical softoware) project lately. But it turned out that the customer changed his mind and decided to go ahead with C# instead. In our company we have .NET and C# Technical Architects, but the management decided that I will be Technical Architect for this project anyway.

So Java guy architecting a C# application. Nice.

As a part of evaluating .NET technologies I decided to try Spring.NET. No one in my company has used it before.

Installing Spring.NET

Installation is very straight forward. Download and execute the msi file (or was it exe?).

First project

I created a Console Application. Then I added references to the following DLLs: Spring.Core.dll and Spring.Aop.dll.

Then I pointed my browser to Spring.NET documentation web site. I found quite a few quick start articles. I started with the first one.

Dependency Injection

Dependency injection works without any problems. Because I was already familiar with Spring framework I just scanned the documentation and I did a small mistake. In my case IMovieFinder has a reference to IMovieListener. In official documentation it's the oposite. Anyway it's just an exercise. My configuration looked like this:

<?xml version="1.0" encoding="utf-8" ?>
<configuration>
    <configSections>
        <sectionGroup name="spring">
            <section name="context" type="Spring.Context.Support.ContextHandler, Spring.Core"/>
            <section name="objects" type="Spring.Context.Support.DefaultSectionHandler, Spring.Core" />
        </sectionGroup>
    </configSections>
    <spring>
        <context>
            <resource uri="config://spring/objects"/>
        </context>
        <objects xmlns="http://www.springframework.net">
            <description>An  example that demonstrates simple IoC features.</description>
            <object id="MyMovieFinder" type="ConsoleApplication1.MovieFinder">
                <property name="Lister" ref="MyMovieLister" />
            </object>
            <object id="MyMovieLister" type="ConsoleApplication1.MovieLister" />
        </objects>
    </spring>
</configuration>

Logging

Next thing I wanted to add was logging. It turned out that in .NET there is Common.Logging framework (architected after Apache commons-logging) which is a bride/abstraction for different logging frameworks. I tried to bridge it with Log4Net, but failed. Log4Net has a dependency to System.Web (sic!). I had to stay with default simple logger offered by Commmon.Logging.

I added Common.Logging.dll to my project. Then I added Common.Logging configuration to my App.config:

<?xml version="1.0" encoding="utf-8" ?>
<configuration>
    <configSections>
        <sectionGroup name="spring">
            <section name="context" type="Spring.Context.Support.ContextHandler, Spring.Core"/>
            <section name="objects" type="Spring.Context.Support.DefaultSectionHandler, Spring.Core" />
        </sectionGroup>
        <sectionGroup name="common">
            <section name="logging" type="Common.Logging.ConfigurationSectionHandler, Common.Logging" />
        </sectionGroup>
    </configSections>
    <spring>
        <context>
            <resource uri="config://spring/objects"/>
        </context>
        <objects xmlns="http://www.springframework.net">
            <description>An  example that demonstrates simple IoC features.</description>
            <object id="MyMovieFinder" type="ConsoleApplication1.MovieFinder">
                <property name="Lister" ref="MyMovieLister" />
            </object>
            <object id="MyMovieLister" type="ConsoleApplication1.MovieLister" />
        </objects>
    </spring>
    <common>
        <logging>
            <factoryAdapter type="Common.Logging.Simple.ConsoleOutLoggerFactoryAdapter, Common.Logging">
                <arg key="level" value="INFO" />
                <arg key="showLogName" value="true" />
                <arg key="showDataTime" value="true" />
                <arg key="dateTimeFormat" value="yyyy/MM/dd HH:mm:ss:fff" />
            </factoryAdapter>
        </logging>
    </common>
</configuration>

Aspect Oriented Programming

My next goal was to create an aspect that would log invocations of my business methods.

Spring.Aop is based on AopAlliance so the concepts and API are almost the same as in Java. I wrote LoggingAroundAdvice:

using System.Text;
using AopAlliance.Intercept;
using Common.Logging;
namespace ConsoleApplication1
{
    public class LoggingAroundAdvice : IMethodInterceptor
    {
        public object Invoke(IMethodInvocation invocation)
        {
            ILog log = LogManager.GetLogger(invocation.Target.GetType());
            StringBuilder sb = new StringBuilder();
            if (invocation.Arguments != null)
            {
                foreach (object arg in invocation.Arguments)
                {
                    sb.Append(arg + " ");
                }
            }
            log.Info("Invoking " + invocation.Method.Name + " with following args => " + sb.ToString());
            object result = invocation.Proceed();
            log.Info("Finished invoking " + invocation.Method.Name + " with result ==> " + result);
            return result;
        }
    }
}

OK. Then I had to modify my App.config. Instead of directly creating my classes I had to create ProxyFactoryObject whose target property was set to the original class. The configuration looked like this:

<?xml version="1.0" encoding="utf-8" ?>
<configuration>
    <configSections>
        <sectionGroup name="spring">
            <section name="context" type="Spring.Context.Support.ContextHandler, Spring.Core"/>
            <section name="objects" type="Spring.Context.Support.DefaultSectionHandler, Spring.Core" />
        </sectionGroup>
        <sectionGroup name="common">
            <section name="logging" type="Common.Logging.ConfigurationSectionHandler, Common.Logging" />
        </sectionGroup>
    </configSections>
    <spring>
        <context>
            <resource uri="config://spring/objects"/>
        </context>
        <objects xmlns="http://www.springframework.net">
            <description>An  example that demonstrates simple IoC features.</description>
            <object id="MyMovieLister" type="Spring.Aop.Framework.ProxyFactoryObject">
                <property name="target">
                    <object id="MyMovieListerTarget" type="ConsoleApplication1.MovieLister, ConsoleApplication1"/>
                </property>
                <property name="interceptorNames">
                    <list>
                        <value>LoggingAroundAdvice</value>
                    </list>
                </property>
            </object>
            <object id="MyMovieFinder" type="Spring.Aop.Framework.ProxyFactoryObject">
                <property name="target">
                    <object id="MyMovieFinderTarget" type="ConsoleApplication1.MovieFinder, ConsoleApplication1">
                        <property name="Lister" ref="MyMovieLister"/>
                    </object>
                </property>
                <property name="interceptorNames">
                    <list>
                        <value>LoggingAroundAdvice</value>
                    </list>
                </property>
            </object>
            <object id="LoggingAroundAdvice" type="ConsoleApplication1.LoggingAroundAdvice, ConsoleApplication1"/>
        </objects>
    </spring>
    <common>
        <logging>
            <factoryAdapter type="Common.Logging.Simple.ConsoleOutLoggerFactoryAdapter, Common.Logging">
                <arg key="level" value="INFO" />
                <arg key="showLogName" value="true" />
                <arg key="showDataTime" value="true" />
                <arg key="dateTimeFormat" value="yyyy/MM/dd HH:mm:ss:fff" />
            </factoryAdapter>
        </logging>
    </common>
</configuration>

But there was one problem with it. The XML configuration was blown out. Number of lines increased and the config was less human readable. I googled for a few seconds and found a better solution: ObjectNameAutoProxyCreator. I updated my App.config one more time:

<?xml version="1.0" encoding="utf-8" ?>
<configuration>
    <configSections>
        <sectionGroup name="spring">
            <section name="context" type="Spring.Context.Support.ContextHandler, Spring.Core"/>
            <section name="objects" type="Spring.Context.Support.DefaultSectionHandler, Spring.Core" />
        </sectionGroup>
        <sectionGroup name="common">
            <section name="logging" type="Common.Logging.ConfigurationSectionHandler, Common.Logging" />
        </sectionGroup>
    </configSections>
    <spring>
        <context>
            <resource uri="config://spring/objects"/>
        </context>
        <objects xmlns="http://www.springframework.net">
            <description>An  example that demonstrates simple IoC features.</description>
            <object id="MyMovieFinder" type="ConsoleApplication1.MovieFinder">
                <property name="Lister" ref="MyMovieLister" />
            </object>
            <object id="MyMovieLister" type="ConsoleApplication1.MovieLister" />
            <object id="LoggingAroundAdvice" type="ConsoleApplication1.LoggingAroundAdvice, ConsoleApplication1"/>
            <object type="Spring.Aop.Framework.AutoProxy.ObjectNameAutoProxyCreator, Spring.Aop">
                <property name="ObjectNames">
                    <list>
                        <value>*Movie*</value>
                    </list>
                </property>
                <property name="InterceptorNames">
                    <list>
                        <value>LoggingAroundAdvice</value>
                    </list>
                </property>
            </object>
        </objects>
    </spring>
    <common>
        <logging>
            <factoryAdapter type="Common.Logging.Simple.ConsoleOutLoggerFactoryAdapter, Common.Logging">
                <arg key="level" value="INFO" />
                <arg key="showLogName" value="true" />
                <arg key="showDataTime" value="true" />
                <arg key="dateTimeFormat" value="yyyy/MM/dd HH:mm:ss:fff" />
            </factoryAdapter>
        </logging>
    </common>
</configuration>

Much, much better. While googing I found out that there is already a logging aspect (which was not a surprise for me). To use it I just added the following object configuration:

<object name="SpringLoggingAroundAdvice" type="Spring.Aspects.Logging.SimpleLoggingAdvice, Spring.Aop">
    <property name="LogUniqueIdentifier" value="false"/>
    <property name="LogExecutionTime"    value="true"/>
    <property name="LogMethodArguments"  value="true"/>
    <property name="LogReturnValue"      value="true"/>
    <property name="Separator"           value=";"/>
    <property name="LogLevel"            value="Info"/>
    <property name="HideProxyTypeNames"  value="true"/>
    <property name="UseDynamicLogger"    value="true"/>
</object>

And changed InterceptorNames property of the ObjectNameAutoProxyCreator object to SpringLoggingAroundAdvice. Worked like a charm.

Internationalisation

OK. Medical software has to support internationalisation. Spring.NET supports resx files. I created Common.resx and Common.pl-PL.resx files. I added definition for "Yes" word. Next I had to add two more objects to my App.config:

<object name="messageSource" type="Spring.Context.Support.ResourceSetMessageSource, Spring.Core">
    <property name="resourceManagers">
        <list>
            <ref object="myResourceManager"/>
        </list>
    </property>
</object>
<object name="myResourceManager" type="Spring.Objects.Factory.Config.ResourceManagerFactoryObject, Spring.Core">
    <property name="baseName">
        <value>ConsoleApplication1.Resources.Common</value>
    </property>
    <property name="assemblyName">
        <value>ConsoleApplication1</value>
    </property>
</object>

Too much lines of code! I found out much shorter solution (in a matter of fact you can mix both approaches):

<object name="messageSource" type="Spring.Context.Support.ResourceSetMessageSource, Spring.Core">
    <property name="resourceManagers">
        <list>
            <value>ConsoleApplication1.Resources.Common, ConsoleApplication1</value>
        </list>
    </property>
</object>

Much better.

In my application's code I wrote:

string msg = ctx.GetMessage(Resources.Common.Yes, CultureInfo.CurrentCulture);
string msgPl = ctx.GetMessage(Resources.Common.Yes, new CultureInfo("pl-PL"));

Ran the application and it worked.

Summary

Not too bad. Not too bad at all for a .NET application :)

Stay tuned, I plan to write some NUnit tests for my Spring.NET project.

cheers,
Łukasz

TeamCity - my first impressions

2011-07-29T10:43:00.000+02:00

So far I have been using CruiseControl as my CI server for both Java and .NET projects. This week I decided to give TeamCity a try.

Installation

The installation is straight forward. For the Windows version you just download the installer and it takes care of everything else (except for pressing Next buttons).

Once the installation is complete, TeamCity service starts and you can log in into TeamCity console.

Setting up project

I have to say that for first 5 minutes I was a little bit muddled. I thought I would make it without reading the documentation. I haven't used it before, but within 15 minutes I was able to find all information I needed from TeamCity web site which is a great plus. Now that I have configured 3 projects in TeamCity I can say that the configuration is intuitive.

In total (excluding time required to install SQL Server, SQL Server SP 2, SQL Server Management Studio, and VS 2010 - unfortunatelly we write tests in MSTest which use Microsoft.VisualStudio namespace) it took me 15 minutes to configure the following things:

1) Build project - 1st build step
2) Run MSTest tests - 2nd build step
3) Run code coverage analysis (TeamCity 6.x is shipped with dotCover) - part of 2nd build step

While I was configuring my 2nd project it took me even less, aprox 5 minutes to configure:

1) Build project
2) Run MSTest tests
3) Run code coverage analysis
4) Update email notification settings

Nice.

TeamCity for C++ developers

Yes. As a part of preparation to my next project I started playing around with C++ and it's support in VS. TeamCity can build C++ projects as Visual Studio Solutions. C++ unit tests are executed by MSTest tool. Source code coverage... works like a charm! I still have to find some kind of StyleCop for C++ and hook it into TeamCity.

TeamCity IDE plugins

TeamCity has many IDE plugins. They can be found on your TeamCity server. In my case I installed VS 2010 plugin (/update/vsAddinInstallerv4.msi) and universal tray notifier app (/update/TrayNotifierInstaller.msi). No problems at all. VS 2010 required restart.

Summary

TeamCity looks like a really professional and powerful tool.

Cheers,
Łukasz

Two way communication in JMS

2011-07-04T10:49:00.001+02:00

Today I will debuke quite a popular myth about one-way only communication in JMS. There is no classic request-response equivalent of course (just like there is in AMQP), but a message may convey a reply to header containing a reference to a temporary queue. This way we can achieve two way communication in JMS.

Below I give you a simple JUnit test which illustrates the idea.

Example

In the following example I use a temporary queue and set it as JMS reply to header. The temporary queue exists as long as its creator's session is active. The code is pretty simple and self commenting. In testA I create a message and a temporary queue which I expect to recieve a response from the consumer. Comsumer in testB consumes the message, retrieves the reply to destination and sends a text message to it. testC consumes the message from temporary queue. Once the session is closed, temporary queue is disposed and removed.

public class AppTest {
 private static Connection connection;
 private static Session session;
 private static Destination destination;
 private static Destination replyToDestination;
 @BeforeClass
 public static void setup() throws JMSException {
  ConnectionFactory connectionFactory = new ActiveMQConnectionFactory(
    "tcp://localhost:62626");
  connection = connectionFactory.createConnection();
  connection.start();
  session = connection.createSession(true, Session.SESSION_TRANSACTED);
  destination = session.createQueue("Java.ActiveMQ.Test.Queue");
  replyToDestination = session.createTemporaryQueue();
 }
 @AfterClass
 public static void cleanup() throws JMSException {
  session.close();
  connection.stop();
  connection.close();
 }
 @Test
 public void testA() throws JMSException {
  MessageProducer producer = session.createProducer(destination);
  Person person = new Person();
  person.setFirstName("Lukasz");
  person.setLastName("Budnik");
  Message message = session.createObjectMessage(person);
  message.setJMSReplyTo(replyToDestination);
  producer.send(message);
  session.commit();
 }
 @Test
 public void testB() throws JMSException, InterruptedException {
  Person person = null;
  MessageConsumer consumer = session.createConsumer(destination);
  Message message;
  while ((message = consumer.receive(2000l)) != null) {
   if (message instanceof ObjectMessage) {
    ObjectMessage objectMessage = (ObjectMessage) message;
    Object object = objectMessage.getObject();
    if (object instanceof Person) {
     person = (Person) object;
     Assert.assertEquals("Lukasz", person.getFirstName());
     Assert.assertEquals("Budnik", person.getLastName());
     Destination replyToDestination = message.getJMSReplyTo();     
     MessageProducer replyToMessageProducer = session.createProducer(replyToDestination);
     Message replyMessage = session.createTextMessage("OK");
     replyToMessageProducer.send(replyMessage);     
     session.commit();
    }
   }
  }
  if (person == null) {
   Assert.fail("Person is null");
  }
 }
 @Test
 public void testC() throws JMSException, InterruptedException {
  String text = null;
  MessageConsumer consumer = session.createConsumer(replyToDestination);
  Message message;
  while ((message = consumer.receive(2000l)) != null) {
   if (message instanceof TextMessage) {
    TextMessage textMessage = (TextMessage) message;
    text = textMessage.getText();
    Assert.assertEquals("OK", text);     
    session.commit();
   }
  }
  if (text == null) {
   Assert.fail("Text is null");
  }
 }
}

Summary

Simple, isn't it? This mechanism is used heavily in Apache Camel and WS frameworks which use transacted JMS under the hood for WS-RM implementations. Using those frameworks, temporary queues and reply to headers are being taken care of transparently so I may even not know that you use them :)

cheers,
Łukasz

ActiveMQ and .NET combined!

2011-06-28T10:58:00.004+02:00

ActiveMQ is one of the most popular messaging frameworks. For sure the most popular open source framework. Many people think that ActiveMQ works only with Java and this is not true at all. ActiveMQ can work with almost every popular language (including JavaScript!) through numerous protocols which it supports.

Today I will show you how to use ActiveMQ in .NET-based solutions.

Project setup

Using VS 2010's Extension Manger I installed NuGet Package Manager. After installation and VS 2010 restart, I created a project called ActiveMQNMS. I right-clicked it and selected "Manage NuGet packages...". In the search field I typed: "ActiveMQ". There was a package called Apache.NMS.ActiveMQ. I installed it. (Note: ActiveMQ has one dependency - Apache.NMS package. The NMS package provides a unified API for working with different messaging frameworks and providers.)

Starting ActiveMQ

I already had ActiveMQ installed on my machine. If you don't have one, download it from http://activemq.apache.org. The default instance listens on 61616 port. However, mine is listening on 62626. If you want to run my code, please remember to change the port.

To start ActiveMQ I executed:

activemq-5.5.0\bin\activemq

Depending on configured ports, you can use ActiveMQ web console to manage your queues, topics, subscribers, connections, embedded Apache Camel, etc. I'm using 8282 port, and the console URL is: http://localhost:8282/admin.

Test stub

In general the .NET API is almost a copy of the Java API. So if you're familiar with JMS and/or ActiveMQ you don't need any documentation. Please note TestIntialize and TestCleanup methods.

using System;
using Apache.NMS;
using Apache.NMS.ActiveMQ;
using Microsoft.VisualStudio.TestTools.UnitTesting;

namespace ActiveMQNMS
{
    [Serializable]
    public class Person
    {
        public string FirstName
        {
            get;
            set;
        }
        public string LastName
        {
            get; set;
        }
    }
    [TestClass]
    public class ActiveMqTest
    {
        private IConnection _connection;
        private ISession _session;
        private const String QUEUE_DESTINATION = "DotNet.ActiveMQ.Test.Queue";
        [TestInitialize]
        public void TestInitialize()
        {
            IConnectionFactory factory = new ConnectionFactory("tcp://localhost:62626");
            _connection = factory.CreateConnection();
            _connection.Start();
            _session = _connection.CreateSession();
        }
        [TestCleanup]
        public void TestCleanup()
        {
            _session.Close();
            _connection.Close();
        }
    }
}

Writing Producer

Here is the producer:

[TestMethod]
public void TestA()
{
    IDestination dest = _session.GetQueue(QUEUE_DESTINATION);
    using (IMessageProducer producer = _session.CreateProducer(dest))
    {
        var person = new Person
        {
            FirstName = "Łukasz",
            LastName = "Budnik"
        };
        var objectMessage = producer.CreateObjectMessage(person);
        producer.Send(objectMessage);
    }
}

Run the test and refresh "Queues" list in ActiveMQ web console. You should see DotNet.ActiveMQ.Test.Queue queue with 1 enqueued and pending message. Purge the queue by hitting the purge link or you simply delete it.

Writing Consumer

Now we have to consume the message. Here is the code:

[TestMethod]
public void TestB()
{
    Person person = null;
    IDestination dest = _session.GetQueue(QUEUE_DESTINATION);
    using (IMessageConsumer consumer = _session.CreateConsumer(dest))
    {
        IMessage message;
        while ((message = consumer.Receive(TimeSpan.FromMilliseconds(2000))) != null)
        {
            var objectMessage = message as IObjectMessage;
            if (objectMessage != null)
            {
                person = objectMessage.Body as Person;
                if (person != null)
                {
                    Assert.AreEqual("Łukasz", person.FirstName);
                    Assert.AreEqual("Budnik", person.LastName);
                }
            }
            else
            {
                Assert.Fail("Object Message is null");
            }
        }
    }
    if (person == null)
    {
        Assert.Fail("Person object is null");
    }
}

Run tests. Refresh "Queues" tab in ActiveMQ web console. You should see 1 message enqueued and 1 message dequeued. As expected.

Summary

That's all. Simple, isn't it? ActiveMQ works very, very nicely with .NET.

I have to find some performance comparison for ActiveMQ and MS or pure .C#/NET messaging frameworks. Or maybe you have it? Please share.

cheers,
Łukasz

Passing backend data to JavaScript

2011-06-20T10:24:00.000+02:00

When you create your front-ends dynamically using JavaScript frameworks you have to work with the backend data: constants, URLs, labels, or localised labels.

Properties files are made for this. The problem is when you use frameworks like SenchTouch and your HTML body looks like this: <body />. The other problem is when some of the data is actually stored in database.

A very bad idea is to hard code copies of these values in JavaScripts. The presentation layer should not store any of these.

In general there are 3 ways of handling such situations. Here they are.

1. Generating JavaScripts directly in views

Here is an example of what I'm talking about:

<script type="text/javascript">
var App.Resources.OKButtonLabel = 'Depending on technology of your choice read the properties here';
</script>

Advantages:

JavaScript is generated when web page is rendered
it's fast

Disadvantages:

not the most elegant way
static - generated only during the render phase
uses only properties files

2. Generating JavaScripts on the fly

A sample:

<script type="text/javascript" src="/dynamic/messages.js"></script>

And the contents of given URL are generated dynamically.

Advantages:

more elegant than generating JavaScripts directly in views
more flexible: can return data stored in properties files as well as data stored in database

Disadvantages:

browser has to download additional JavaScripts
static - contents of external JavaScripts are loaded only onced

3. Using AJAX

Advantages:

fully dynamic behaviour
data can be reloaded at any time
supports dynamic localisation and language change
can return data stored in properties files as well as data stored in database

Disadvantages:

asynchronous calls are slower by nature
additional JavaScript logic is required

Summary

All 3 of the above approaches work, but when designing your application you have to balance all pros and cons and choose the one that suits your needs.

cheers,
Łukasz

Transactions and Isolation Levels explained once and for all (nHibernate examples)

2011-06-02T11:53:00.001+02:00

A question about transactions and isolation levels is one of my favourites when interviewing software engineers. Some of them heard about it, some of them know what they are, some of them can list them, some of them can describe them in detail. But there are also developers that don't have the slightest idea what are isolation levels.

Today, as a part of my nHibernate series I'll show you a unit test which shows you the behaviour of 4 isolation levels: read uncommitted, read committed, repeatable read, and serialisable.

The project

For this excercise I used nHibernate project I wrote last time. The post is here: Playing around with nHibernate: the basics.

Test structure

OK. To test concurent accesses I had to create 2 independent sessions and 2 independent transactions. I create sessions and transactions in [TestInitialize] and dispose them in [TestCleanup]. Just in case something goes wrong I delete all records from app_country in [ClassCleanup] method.

The structure looks like this:

[TestClass]
public class TransactionsTest
{

    private ISession SessionA { get; set; }
    private ISession SessionB { get; set; }
    private ITransaction TransactionA { get; set; }
    private ITransaction TransactionB { get; set; }

    [TestInitialize]
    public void Startup()
    {
        SessionA = OpenSession();
        SessionB = OpenSession();
        TransactionA = SessionA.BeginTransaction();
        TransactionB = SessionB.BeginTransaction();
    }

    [TestCleanup]
    public void Cleanup()
    {
        TransactionA.Dispose();
        TransactionB.Dispose();
        SessionA.Dispose();
        SessionB.Dispose();
    }

    [ClassCleanup]
    public static void ClassCleanup()
    {
        using (ISession session = OpenSession())
        {
            using (ITransaction transaction = session.BeginTransaction())
            {
                IQuery query = session.CreateSQLQuery("delete from app_country");
                query.ExecuteUpdate();
                transaction.Commit();
            }
        }
    }

    private static ISessionFactory _sessionFactory;
    private static ISession OpenSession()
    {
        if (_sessionFactory == null)
        {
            Configuration configuration = new Configuration();
            configuration.AddAssembly(Assembly.GetCallingAssembly());
            _sessionFactory = configuration.BuildSessionFactory();
        }
        return _sessionFactory.OpenSession();
    }
}

Rolling back transactions

Before I show you the isolation levels I'd like to show you how to rollback a transaction. The code is very simple and self commenting:

[TestMethod]
public void TestRollback()
{
    Country country = new Country()
    {
        Description = "Poland",
        ContinentId = 1
    };

    try
    {
        SessionA.Save(country);
        Assert.IsTrue(country.Id > 0);
        TransactionA.Rollback();
        Assert.IsTrue(TransactionA.WasRolledBack);
        IQuery query = SessionA.CreateQuery("FROM Country WHERE Id = :id");
        query.SetInt32("id", country.Id);
        Country nonExistingRolledBackCountry = query.UniqueResult<Country>();
        Assert.IsNull(nonExistingRolledBackCountry);
    }
    catch (Exception e)
    {
        Assert.Fail("Test failed " + e.Message);
    }
}

Default Isolation Level

OK. When you create a transaction using just BeginTransaction() (without the isolation level parameter) it starts the transaction in default isolcation level. If you don't know what is the default isolation level you will know it after reading this post :)

This and all next tests use SessionA and SessionB objects to simulate 2 concurent clients.

public void TestDefaultIsolationLevel()
{
    Country countryA = new Country()
    {
        Description = "Poland",
        ContinentId = 1
    };

    try
    {
        // save an object in session A
        SessionA.Save(countryA);
        Assert.IsTrue(countryA.Id > 0);

        // try to retrieve newly inserted object from session B
        IQuery query = SessionB.CreateQuery("FROM Country WHERE Id = :id");
        query.SetInt32("id", countryA.Id);
        Country countryB = query.UniqueResult<Country>();
    }
    catch (GenericADOException ex)
    {
        // this is the expected behaviour!
        Assert.AreSame(typeof(GenericADOException), ex.GetType());
    }
    catch (Exception ex)
    {
        Assert.Fail("Unexpected exception " + ex.Message);
    }
    finally
    {
        TransactionA.Rollback();
    }
}

Result: Session B client is not able to retrieve the country because it's not yet committed. The root cause is timeout exception.

Read Uncommitted Isolation Level

This is the lowest isolation level. In this isolation level you can retrieve uncommitted records:

[TestMethod]
public void TestReadUncommittedIsolationLevel()
{
    Country countryA = new Country()
    {
        Description = "Poland",
        ContinentId = 1
    };

    // close session B transaction
    TransactionB.Dispose();
    // create a new one with ReadUncommitted isolation level
    TransactionB = SessionB.BeginTransaction(System.Data.IsolationLevel.ReadUncommitted);

    try
    {
        // save an object in session A
        SessionA.Save(countryA);
        Assert.IsTrue(countryA.Id > 0);

        // try to retrieve newly inserted object from session B
        IQuery query = SessionB.CreateQuery("FROM Country WHERE Id = :id");
        query.SetInt32("id", countryA.Id);
        Country countryB = query.UniqueResult<Country>();
        Assert.IsNotNull(countryB);

        // rollback transaction A
        TransactionA.Rollback();
        // session B stays with dirty record countryB
    }
    catch (Exception ex)
    {
        Assert.Fail("Unexpected exception " + ex.Message);
    }
}

Result: Session B reads dirty (uncommitted) record, session A rolls back the transaction, Session B stays with dirty record.

Read Committed Isolation Level

This isolation level guarantees no dirty records.

[TestMethod]
public void TestReadCommittedIsolationLevel()
{
    Country countryA = new Country()
    {
        Description = "Poland",
        ContinentId = 1
    };

    TransactionB.Dispose();
    // create new transaction
    TransactionB = SessionB.BeginTransaction(System.Data.IsolationLevel.ReadCommitted);

    try
    {
        // save an object in session A
        SessionA.Save(countryA);
        Assert.IsTrue(countryA.Id > 0);

        // try to retrieve newly inserted object from session B
        IQuery query = SessionB.CreateQuery("FROM Country WHERE Id = :id");
        query.SetInt32("id", countryA.Id);
        Country countryB = query.UniqueResult<Country>();
    }
    catch (GenericADOException ex)
    {
        // this is expected behaviour!
        Assert.AreSame(typeof(GenericADOException), ex.GetType());
    }
    catch (Exception ex)
    {
        Assert.Fail("Unexpected exception " + ex.Message);
    }
    finally
    {
        TransactionA.Rollback();
    }
}

Result: Session B cannot read uncommitted record.

Repeatable Read Isolation Level

This isolation level guarantees repeatable reads. If a repeatable read isolation level is used, a read lock is acquired on fetched records.

[TestMethod]
public void TestRepeatableReadIsolationLevel()
{
    Country countryA = new Country()
    {
        Description = "Poland",
        ContinentId = 1
    };
    Country countryB = null;

    // close current session B transaction
    TransactionB.Dispose();
    // open new transaction with RepeatableRead isolation level
    TransactionB = SessionB.BeginTransaction(System.Data.IsolationLevel.RepeatableRead);

    try
    {
        // save an object in session A
        SessionA.Save(countryA);
        Assert.IsTrue(countryA.Id > 0);

        TransactionA.Commit();
        // open new transaction
        TransactionA.Dispose();
        TransactionA = SessionA.BeginTransaction();
        // retrieve from session A
        IQuery queryA = SessionA.CreateQuery("FROM Country WHERE Id = :id");
        queryA.SetInt32("id", countryA.Id);
        countryA = queryA.UniqueResult<Country>();

        // retrieve from session B
        IQuery queryB = SessionB.CreateQuery("FROM Country WHERE Id = :id");
        queryB.SetInt32("id", countryA.Id);
        countryB = queryB.UniqueResult<Country>();

        SessionA.Delete(countryA);
        TransactionA.Commit();
    }
    catch (GenericADOException ex)
    {
        // this is the expected behaviour!
        Assert.AreSame(typeof(GenericADOException), ex.GetType());
    }
    catch (Exception ex)
    {
        Assert.Fail("Unexpected exception " + ex.Message);
    }
    finally
    {
        SessionB.Delete(countryB);
        TransactionB.Commit();
    }
}

Result: Session A cannot update/delete country, because session B acquired a read lock thanks to using a transaction with a repeatable read isolation level.

Some databases don't acquire read locks (called range-locks) when fetching large data sets like where Id between (0, 1000000). Thus phantom records can occur.

Serialisable Isolation Level

This is the most resource-consuming (and thus slowest) transaction level. But it guarantees a complete transaction isolation. All transactions are serialisable. No phantom records occur.

[TestMethod]
public void TestSerialisable()
{
    Country countryA = new Country()
    {
        Description = "Poland",
        ContinentId = 1
    };

    // close current session B transaction
    TransactionB.Dispose();
    // open new transaction with Serializable isolation level
    TransactionB = SessionB.BeginTransaction(System.Data.IsolationLevel.Serializable);

    try
    {
        // save an object in session A
        SessionA.Save(countryA);
        Assert.IsTrue(countryA.Id > 0);

        TransactionA.Commit();
        // open new transaction
        TransactionA.Dispose();
        TransactionA = SessionA.BeginTransaction();
        // retrieve in 1 session
        IQuery queryA = SessionA.CreateQuery("FROM Country WHERE Id = :id");
        queryA.SetInt32("id", countryA.Id);
        countryA = queryA.UniqueResult<Country>();

        // if I would like to delete a record there would be a timeout exception see TestRepeatableReadIsolationLevel()
        // alghought I haven't used any INSERT/UPDATE/DELETE statements I have to release read locks
        // if I close the transaction B, transaction A will work
        TransactionB.Rollback();

        SessionA.Delete(countryA);
        TransactionA.Commit();
    }
    catch (GenericADOException ex)
    {
        // this is expected behaviour!
        Assert.AreSame(typeof(GenericADOException), ex.GetType());
    }
    catch (Exception ex)
    {
        Assert.Fail("Unexpected exception " + ex.Message);
    }
}

Summary

So, what is the default transaction isolation level?

cheers,
Łukasz

Playing around with nHibernate: the basics

2011-05-30T11:08:00.000+02:00

I used Hibernate in a quite a few Java projects. Last week I was asked by my .NET colleagues to show them nHibernate.

I created a series of simple tests showing them some of nHibernate features. Today I publish the first one.

The basics of Hibernate: Session, CRUD operations, and Criteria API.

Database

OK. For a start just 1 table.

I opened my SQL Express Management Studio and created a database tech_clash_test_db. Then I created app_country table:

CREATE TABLE [dbo].[app_country](
 [countryId] [int] IDENTITY(1,1) NOT NULL,
 [description] [nvarchar](50) NOT NULL,
 [continentId] [int] NOT NULL
);

Setting up the project

First, I had to download Hibernate and add NHibernate and Iesi.Collections (shipped with nHibernate) references to my project.

Just like in Java, you can configure nHibernate in two ways. By using XML descriptors or by using attributes (Java annotations). Today I will show you the first one.

For the given app_country table I created Country.hbm.xml file (Embedded Resource) and wrote:

<?xml version="1.0" encoding="utf-8" ?>
<hibernate-mapping xmlns="urn:nhibernate-mapping-2.2" auto-import="true">
    <class name="nHibernateTest.Model.Country, nHibernateTest" table="app_country">
        <id name="Id" column="countryId">
            <generator class="native" />
        </id>
        <property name="Description" column="description"/>
        <property name="ContinentId" column="continentId"/>
    </class>
</hibernate-mapping>

Having a table and its nHibernate mapping I only had to create a model class (yes, properties are virtual):

public class Country
{
    public virtual int Id { get; set; }

    public virtual string Description { get; set; }

    public virtual int ContinentId { get; set; }
}

Also, I had to tell my application that I used Hibernate. I created App.config file and copied and pasted the following configuration:

<?xml version="1.0" encoding="utf-8" ?>
<configuration>
    <configSections>
        <section name="hibernate-configuration" type="NHibernate.Cfg.ConfigurationSectionHandler, NHibernate" />
        <section name="log4net" type="log4net.Config.Log4NetConfigurationSectionHandler, log4net" />
    </configSections>
    <hibernate-configuration xmlns="urn:nhibernate-configuration-2.2">
        <session-factory>
            <property name="connection.provider">
                NHibernate.Connection.DriverConnectionProvider
            </property>
            <property name="connection.driver_class">
                NHibernate.Driver.SqlClientDriver
            </property>
            <property name="connection.connection_string">
                Server=.\SQLEXPRESS1;Database=tech_clash_test_db;Trusted_Connection=True;MultipleActiveResultSets=True
            </property>
            <property name="dialect">
                NHibernate.Dialect.MsSql2000Dialect
            </property>
            <property name="show_sql">
                true
            </property>
        </session-factory>
    </hibernate-configuration>
    <log4net>
        <appender name="ConsoleAppender"
        type="log4net.Appender.ConsoleAppender, log4net">
            <layout type="log4net.Layout.PatternLayout, log4net">
                <param name="ConversionPattern" value="%m" />
            </layout>
        </appender>
        <root>
            <priority value="WARN" />
            <appender-ref ref="ConsoleAppender" />
        </root>
    </log4net>
</configuration>

That's it.

Writing unit test

Using VS 2010 I created a Test Project. I changed UnitTest1.cs to CRUDTest.cs and wrote:

using System;
using System.Reflection;
using System.Collections.Generic;
using Microsoft.VisualStudio.TestTools.UnitTesting;
using NHibernate;
using NHibernate.Cfg;
using NHibernate.Criterion;
using nHibernateTest.Model;

namespace nHibernateTest
{
    [TestClass]
    public class CRUDTest
    {
        public static int CountryId { set; get; }

        public TestContext TestContext { set; get; }

        [ClassCleanup()]
        public static void ClassCleanup()
        {
            using (ISession session = OpenSession())
            {
                using (ITransaction transaction = session.BeginTransaction())
                {
                    IQuery query = session.CreateSQLQuery("delete from app_country");
                    query.ExecuteUpdate();
                    transaction.Commit();
                }
            }   
        }

        [TestMethod]
        public void TestCreate()
        {
            Country country = new Country()
            {
                Description = "Poland",
                ContinentId = 1
            };

            try
            {
                using (ISession session = OpenSession())
                {
                    using (ITransaction transaction = session.BeginTransaction())
                    {
                        session.Save(country);
                        transaction.Commit();
                        Assert.IsTrue(country.Id > 0);
                        CountryId = country.Id;
                    }
                }
            }
            catch (Exception e)
            {
                Assert.Fail("Test failed " + e.Message);
            }
        }

        [TestMethod]
        public void TestRetrieveAll()
        {
            try
            {
                using (ISession session = OpenSession())
                {
                    IQuery query = session.CreateQuery("SELECT c FROM Country c");
                    IList<Country> countries = query.List<Country>();
                    Assert.IsNotNull(countries);
                    Assert.IsTrue(countries.Count > 0);
                }
            } catch (Exception e)
            {
                Assert.Fail("Test failed " + e.Message);
            }
        }

        [TestMethod]
        public void TestRetrieveSingleRecord()
        {
            try
            {
                using (ISession session = OpenSession())
                {
                    using (ITransaction transaction = session.BeginTransaction())
                    {
                        IQuery query = session.CreateQuery("FROM Country WHERE Id = :id and Description = :description ORDER BY Description ASC");
                        query.SetInt32("id", CountryId);
                        query.SetString("description", "Poland");
                        Country country = query.UniqueResult<Country>();
                        Assert.AreEqual("Poland", country.Description);
                    }
                }
            }
            catch (Exception e)
            {
                Assert.Fail("Test failed " + e.Message);
            }
        }

        [TestMethod]
        public void TestUpdate()
        {
            try
            {
                using (ISession session = OpenSession())
                {
                    using (ITransaction transaction = session.BeginTransaction())
                    {
                        IQuery query = session.CreateQuery("FROM Country WHERE Id = :id");
                        query.SetInt32("id", CountryId);
                        Country country = query.UniqueResult<Country>();
                        country.Description = "Poland 2";
                        transaction.Commit();
                        Country country2 = query.UniqueResult<Country>();
                        Assert.AreEqual("Poland 2", country2.Description);
                    }
                }
            }
            catch (Exception e)
            {
                Assert.Fail("Test failed " + e.Message);
            }
        }

        [TestMethod]
        public void TestDelete()
        {
            try
            {
                using (ISession session = OpenSession())
                {
                    using (ITransaction transaction = session.BeginTransaction())
                    {
                        IQuery query = session.CreateQuery("FROM Country WHERE Id = :id");
                        query.SetInt32("id", CountryId);
                        Country country = query.UniqueResult<Country>();
                        session.Delete(country);
                        transaction.Commit();
                        Country country2 = query.UniqueResult<Country>();
                        Assert.IsNull(country2);
                    }
                }
            }
            catch (Exception e)
            {
                Assert.Fail("Test failed " + e.Message);
            }
        }

        [TestMethod]
        public void TestCriteriaAPI()
        {
            try
            {
                using (ISession session = OpenSession())
                {
                    using (ITransaction transaction = session.BeginTransaction())
                    {
                        ICriteria criteria = session.CreateCriteria(typeof(Country));
                        criteria.Add(Restrictions.Eq("Description", "Poland"));
                        criteria.Add(Restrictions.Eq("Id", CountryId));
                        criteria.AddOrder(Order.Asc("Description"));

                        Country country = criteria.UniqueResult<Country>();

                        Assert.IsNotNull(country);
                    }
                }
            }
            catch (Exception e)
            {
                Assert.Fail("Test failed " + e.Message);
            }
        }

        private static ISessionFactory _sessionFactory;
        private static ISession OpenSession()
        {
            if (_sessionFactory == null)
            {
                Configuration configuration = new Configuration();
                configuration.AddAssembly(Assembly.GetCallingAssembly());
                _sessionFactory = configuration.BuildSessionFactory();
            }
            return _sessionFactory.OpenSession();
        }
    }
}

OK. I don't think comments are needed here.

As you can see, the core component is the Session component. Session component represents a connection to a database and it allows you to create transactions, queries, criteria, and of course perform CRUD operations.

One more thing. The above test fails. The order of execution is relevant. But that's not a problem. Using VS 2010 I created ordered test. The most important thing is to place TestCreate() method as a first method and TestDelete() as the last one. In my case I ordered them in the following way (the output of ordered test results):

Passed Unit Test TestCreate  00:00:00.0033537
Passed Unit Test TestRetrieveAll  00:00:00.0023378
Passed Unit Test TestRetrieveSingleRecord  00:00:00.0031339
Passed Unit Test TestCriteriaAPI  00:00:00.0037914
Passed Unit Test TestUpdate  00:00:00.0046633
Passed Unit Test TestDelete  00:00:00.0203877

Summary

Next time I will show you either transactions with different isolation levels or I will show you 1:1, 1:n, and n:m mappings with lazy and eager loading.

Stay tuned,
Łukasz

True power of Java EE and JMS: distributed computations!

2011-05-24T12:17:00.000+02:00

As you already know I'm writing a PhD about evaluating and measuring quality of Web Services compositions. In my system I focus on evaluating WS-BPEL processes (but the framework I designed is extremely flexible and supports any business scenario technology).

The system (called Measurement System DIES) is a Java EE application. It is designed in a distributed manner which makes scaling out and distributing computations very easy.

Last week, out of pure curiosity, I decided to do stress tests. Apache Geronimo did not survive this nor did the embedded ActiveMQ. After I killed Geronimo and started it again all queues stopped working... I had to install brand new instance of Geronimo.

But thanks to Java EE capabilities and the distributed nature of my system I was able to overcome this problem. See how I did it with JMS and distributed computations!

The EJB module of my application

Most of the work performed by my system is done, of course, in EJB module. The EJB module of my application consists of the following (among many other things of course):

Queue for execution orders - each execution of a WS-BPEL process instance may last minutes (I have a 10 minutes timeout set up),
Queue for asynchronous computing of quality and performance score of each activity trace - this task usually takes up to 4 seconds to complete (most of the time spent here is to query external services like geolocation services),
Timer for computing statistics - every 10 seconds I retrieve all completed instances which are not yet analysed and conduct some statistical computations.

The development environment

While developing my system I used 1 box for everything:

Apache Geronimo with my Java EE application and embedded ActiveMQ server
Apache ODE as an WS-BPEL execution engine
IBM DB2 pureXML

New Machine(s)

I managed to get a second machine. Based on that I decided to distribute my components this way:

1st computer:

Apache Geronimo with my Java EE application with Timer but without Queue components
IBM DB2 pureXML
Apache ODE as an execution engine

2nd computer:

Apache Geronimo with EJB module with Queue components
Standalone ActiveMQ

Ideally I would need 2 more machines:

3rd for IBM DB2 pureXML database
4th for Apache ODE

But having 2 machines was enough.

Distributing components

Here is what I did:

I split my EJB components. I moved my two queues into a separate project and deployed them to Apache Geronimo installed on the second machine. I decided to leave the Timer component in the original EJB project.
I downloaded a standalone ActiveMQ server and installed it on the second machine.
Finally I configured connection factories for both Apache Geronimos. I used the wizard offered by Geronimo's Admin Console - it was childishly simple! I use application-wide JMS resources and just compared the generated deployment plans with the ones I already had. It turned out that I had to add 1 line to my deployment plans - the external server url. NICE! This is how the enterprise configuration should look like!

Rerunning the stress test again

I re-ran my stress test. Both Geronimos and ActiveMQ survived :)

What was the improvement rate? Hard to say... In Java when you add, subtract, multiply or divide NaN by any other number you still get NaN. I know one thing for sure, the first configuration did not survive the stress test :)

Conclusions

Always keep in mind that some day your system will have to be distributed or scaled out. At the design stage it's far more easier to introduce basic enterprise architectural patterns and concepts than to introduce them during the maintenance phase (when, for example, task distribution would have a huge impact on the whole system).

cheers,
Łukasz

Java sucks, Groovy and C# rock!

2011-05-16T11:00:00.003+02:00

When I was a student a few years ago one of my lecturers told us that Java and C# can be described by an adjective starting with the "M" letter: Java is Mature and C# is Modern.

To be completely frank with you I think that Java should be described by two adjectives starting with "O": old and obsolete.

Note that I have over 6 years of experience in Java. I'm Sun Certified Java Programmer, Sun Certified Developer for Java Web Services and also Java EE and Spring are still my technologies of choice. I'm just talking about the language here...

Despite that Java as a language sucks very much there is one good thing about it. It's the JVM. JVM is efficient, stable, and above all portable. Atop JVM new dynamic languages are developed. Like for example Groovy and Scala. Last year I wrote a post about 21 things I like and don't like about Groovy (compared to vanilla Java) - you may find it here: 21 things I like and don't like about Groovy.

It would be a low blow if I compared Java to C# (it's like comparing a BMW to a Fiat).

Let's try Groovy with C#.

1. No primitives

In Groovy everything is an object. In C# int, double, and so on, are structs with methods.

Groovy:

def i = 148;

def test = Integer.class == i.class

C#:

int i = 12;
int j = 12;
bool test = i.Equals(j);
int hash = 148.GetHashCode();

Result: draw

2. Named parameters

Groovy:

def person = new Person(name: "Łukasz Budnik", age: 27);

C#:

Person person = new Person()
{
name = "Łukasz Budnik",
age = 27
};

Result: draw

2. Default parameters

Groovy:

def s = printStars()
assertEquals('***', s)
s = printStars(2)
assertEquals('**', s)

String printStars(count = 3) {
def sb = new StringBuilder()
while(count--) {
sb.append ('*')
}
return sb
}

C#:

string stars1 = PrintStars();
string stars2 = PrintStars(13);

public static string PrintStars(int count = 3)
{
    StringBuilder sb = new StringBuilder();

    while (count-- > 0)
    {
        sb.Append("*");
    }

    return sb.ToString();
}

Result: draw

3. Handling arrays, lists, maps

Groovy:

def array = [1, 2, 3, 4, 4, 3, 2, 1]
def set = [1, 2, 3, 4, 4, 3, 2, 1] as Set

C#:

int[] array = {12,12,12,12};
IEnumerable<int> enumerable = new List<int> { 12, 12, 12, 12 };
IEnumerable<int> set = new HashSet<int> { 12, 12, 12, 12 };
IDictionary<string, string> dict = new Dictionary<string, string> { {"key1", "value1"}, { "key2", "value2"} };

Result: draw (slightly Groovy)

4. Operator overloading

Groovy:

class Box {
   def w;
   def h; 
   Box plus(Box other) {
      return new Box(w: this.w + other.w, h: this.h + other.h)
   }
}


def box1 = new Box(w: 100, h: 100)
def box2 = new Box(w: 40, h: 40);
def box3 = box1 + box2;

C#:

class Box
{
   public int W { get; set; }
   public int H { get; set; }

   public static Box operator +(Box b1, Box b2)
   {
     return new Box { W = b1.W + b2.W, H = b1.H + b2.H};
   }
}

var box1 = new Box { W = 100, H = 100 };
var box2 = new Box { W = 40, H = 40 };
var box3 = box1 + box2;

Result: draw

5. Multi-line strings and parsing strings

Multi-line string and parsing ${variable} ${closure(param1)} in string literals in Groovy:

def variable1 = 'I\'m a string'
def variable2 = 123
def label = "${variable1} and my length is not ${variable2}"

and in C#:

var variable1 = @"I'm a
multi-line string";
var variable2 = 123;
var label = String.Format("{0} and my length is not {1}", variable1, variable2);

Result: draw (slightly Groovy)

6. Closoures

A closure in C# takes the form of an in-line delegate method and is strongly typed.

Groovy:

def square = { num -> num * num }
def x = square(10)

C#:

delegate int square(int x);
square s = delegate(int x) { return x * x; };
int squareInt = s(10);

Apart of this in C# you have also Delegate and Func mechanisms. Groovy has also pointers to methods, which is uber cool (see point 10 in this post).

Result: draw

7. Operations on collections

Operations on collections... C# rocks with its LINQ technology (See this for a small sample of what can be done with LINQ: http://msdn.microsoft.com/en-us/vcsharp/aa336746).

If you want to invoke a method on all objects in a collection, in Groovy you would use *. operator:

def x = [ ' aaa ', 'b', ' c  ']
x = x*.trim()

in C# you would use the following LINQ construct:

IEnumerable<string> strings = new List<string> { " asas ", "sdas   ", "           asds  " };
IEnumerable<string> trimmedStrings = strings.Select(it => it.Trim());

Result: C# wins with its LINQ technology (actually C# gets +2 points here)

See the link above to have a full picture of LINQ, another example may be grouping countries list by continents:

def countriesGroupedByContinents = from country in countries
    group country by country.ContinentId into g
    orderby g.Key
    select g;

The SQL-like tree syntax embedded directly into a programming language, what a beauty and a fantastic idea!
(hold on, I, obdurate Java programmer, just wrote that M$ technology is fantastic? Something wrong is happening to me or is it C#...)

8. ?. operator - null-safe invocation, null coalescing operator

Groovy - much more powerful:

def person = new Person(name: 'Łukasz Budnik')
def street = person?.address?.street

C# - only a poor substitute

int? x = null;
int y = x ?? -1;

Result: Groovy wins (back on track, tenacious Java programmer)

9. == operator

Both languages == is more logical/intuitive than in Java, doesn't compare the identities, compares objects.

Groovy:

def name1 = 'Łukasz Budnik'
def name2 = new StringBuilder().append('Łukasz').append(' ')
name2 += 'Budnik'
assertTrue(name1 == name2)
// identity comparison, old Java == behaviour
assertFalse(name1.is(name2))

C#:

var name1 = "Łukasz Budnik";
var name2 = new StringBuilder().Append("Łukasz").Append(" ").Append("Budnik").ToString();

bool test1 = name1 == name2;
// identity comparison
bool test2 = Object.ReferenceEquals(name1, name2);

Result: draw

10. Expando and extensions mechanism

Groovy:

String stringAsStars(def length) {
   def sb = new StringBuilder();
   while (length--) {
      sb.append '*'
   }
   return sb;
}

String.metaClass.addQuestionMark = {->
   return delegate + '?'
}
String.metaClass.addExclamationMark = {->
   return delegate + '!'
}
// ex-C programmer? pointers to functions here? rings any bells?
def method = this.&stringAsStars
String.metaClass.asStars = {->
   return method (delegate.length())
}
String question = 'Are you Groovy programmer'.addQuestionMark()
String answer = 'Yes'.addExclamationMark()
println question
println answer
println answer.asStars()

C#:

StringExtension.AsStarsDelegateFunction = Program.AsStarsImplementation;
string question = "Are you C# programmer".AddQuestionMark();
string answer = "Yes".AddExclamationMark();
string stars = answer.AsStars();

public static string AsStarsImplementation(int length)
{
    StringBuilder sb = new StringBuilder();
    while (length-- > 0)
    {
        sb.Append("*");
    }
    return sb.ToString();
}

public static class StringExtension
{
    public delegate string AsStarsDelegate(int length);

    public static AsStarsDelegate AsStarsDelegateFunction;

    public static string AddQuestionMark(this string s)
    {
        return s + "?";
    }

    public static string AddExclamationMark(this string s)
    {
        return s + "?";
    }
    public static string AsStars(this string s)
    {
        return AsStarsDelegateFunction(s.Length);
    }
}

Result: draw

11. XML/JSON markups

Groovy (More in First steps with Groovy and its XML support post):

def writer = new StringWriter()
def xml = new MarkupBuilder(writer)
xml.people() {
   person(id:123) {
      name ('Lukasz Budnik')
      age(27)
   }
   person(id:321) {
      name ('Budnik Lukasz')
      age(72)
   }
}
def source = writer.toString()
println source

In C# there are anonymous types, but there is no direct XML or JSON builders:

var json = new
{
    name = "Łukasz Budnik",
    id = 123,
    address = new
    {
        city = "Gdansk",
        country = "Poland"
    }
};

But using ExpandoObjects:

dynamic contact = new ExpandoObject();
contact.Name = "Łukasz Budnik";
contact.Address = new ExpandoObject();
contact.Address.City = "Gdansk";
contact.Address.Country = "Poland";

you can write your own wrappers and dynamically format dynamic objects as XML documents:

http://blogs.msdn.com/b/csharpfaq/archive/2009/10/19/dynamic-in-c-4-0-creating-wrappers-with-dynamicobject.aspx :)

Result: draw (dynamic objects C#, markup builders Groovy)

13. Unchecked exceptions

In Groovy and C# all exceptions are unchecked. I personally don't like it. It would be nice to have syntax support, like in Java throws in method declaration, so that developer will know what exceptions might be thrown by given method.

14. Output parameters

Only available in C#.

Result: C# wins

15. Partial classes

Count me in! The second best C# feature (after LINQ). Why/When is it important? Suppose you are generating some of the classes dynamically (base on DB schema) apart of ordinary properties and Equals/HashCode methods you may want to add convenient methods as well - these could be placed in a partial class.

Summary

Do you remember that when Oracle refused to certify Apache Harmony as a Java implementation, there were some voices in Apache Foundation calling for abandoning Java and moving to a different language? Can you imagine Apache CXF, Apache Tomcat, Apache Geronimo, Apache ServiceMix, Apache Camel or Apache Hadoop written in C#? I do!

And what about you? What do you like and don't like about Java and C#?

cheers,
Łukasz

Array Operations in WS-BPEL

2011-04-19T12:59:00.001+02:00

Almost 3 years ago I wrote a short article about handling complex types in WS-BPEL. Today I will show you how to handle arrays.

I used Eclipse BPEL Designer, Apache ODE as a WS-BPEL engine, and soapUI as a Web Services testing tool.

Let's start.

Simple process

First, I created a synchronous process using Eclipse BPEL Designer wizard. I didn't modify the WSDL file I just added the following assign activity to the WS-BPEL stub:

<bpel:assign validate="no" name="Assign">
  <bpel:copy>
   <bpel:from>
    <bpel:literal xml:space="preserve"><tns:ArrayOperationsBusinessProcessResponse
     xmlns:tns="http://jee-bpel-soa.blogspot.com/ws-bpel/array-operations"
     xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
  <tns:result></tns:result>
</tns:ArrayOperationsBusinessProcessResponse>
</bpel:literal>
   </bpel:from>
   <bpel:to variable="output" part="payload"></bpel:to>
  </bpel:copy>
  <bpel:copy>
   <bpel:from>
                    <![CDATA[concat('Hello ', $input.payload/tns:input, '! How are you?')]]>
  </bpel:from>
  <bpel:to part="payload" variable="output">
   <bpel:query queryLanguage="urn:oasis:names:tc:wsbpel:2.0:sublang:xpath1.0"><![CDATA[tns:result]]></bpel:query>
  </bpel:to>
 </bpel:copy>
</bpel:assign>

To invoke it, I used soapUI and the following input message:

<soapenv:Envelope xmlns:soapenv="http://schemas.xmlsoap.org/soap/envelope/" xmlns:arr="http://jee-bpel-soa.blogspot.com/ws-bpel/array-operations">
   <soapenv:Header/>
   <soapenv:Body>
      <arr:ArrayOperationsBusinessProcessRequest>
         <arr:input>Łukasz</arr:input>
      </arr:ArrayOperationsBusinessProcessRequest>
   </soapenv:Body>
</soapenv:Envelope>

The result was:

<soapenv:Envelope xmlns:soapenv="http://schemas.xmlsoap.org/soap/envelope/">
   <soapenv:Body>
      <ArrayOperationsBusinessProcessResponse xmlns="http://jee-bpel-soa.blogspot.com/ws-bpel/array-operations">
         <tns:result xmlns:tns="http://jee-bpel-soa.blogspot.com/ws-bpel/array-operations">Hello Łukasz! How are you?</tns:result>
      </ArrayOperationsBusinessProcessResponse>
   </soapenv:Body>
</soapenv:Envelope>

Adding multiple input elements

OK, I had it working, now it was time to introduce the input array.

All what was required was to add maxOccurs="unbounded" to the WSDL file just like this:

<element name="input" maxOccurs="unbounded" type="string"/>

Now, I was able to invoke it this way:

<soapenv:Envelope xmlns:soapenv="http://schemas.xmlsoap.org/soap/envelope/" xmlns:arr="http://jee-bpel-soa.blogspot.com/ws-bpel/array-operations">
   <soapenv:Header/>
   <soapenv:Body>
      <arr:ArrayOperationsBusinessProcessRequest>
         <arr:input>Łukasz</arr:input>
         <arr:input>Jerzy</arr:input>
         <arr:input>Izydor</arr:input>
         <arr:input>Budnik</arr:input>
      </arr:ArrayOperationsBusinessProcessRequest>
   </soapenv:Body>
</soapenv:Envelope>

But for the above input, the WS-BPEL returned the first input value (Łukasz).

Iterating over an array

When you iterate over an array in WS-BPEL keep in mind the following:

indexes start with 1!!
as a consequence to the above, the end condition is <= (less or equal)
when you index an element in an array, cast the counter value to number using number() function
don't forget to increment the counter otherwise you will end up in infinite loop :)

The updated code was:

<bpel:while name="While">
 <bpel:condition><![CDATA[$counter <= count($input.payload/tns:input)]]></bpel:condition>
 <bpel:sequence>
  <bpel:assign validate="no" name="Assign Output">
   <bpel:copy>
    <bpel:from>
                <![CDATA[concat($output.payload/tns:result, ' ', $input.payload/tns:input[number($counter)])]]>
    </bpel:from>
    <bpel:to part="payload" variable="output">
     <bpel:query queryLanguage="urn:oasis:names:tc:wsbpel:2.0:sublang:xpath1.0"><![CDATA[tns:result]]></bpel:query>
    </bpel:to>
   </bpel:copy>
  </bpel:assign>
 </bpel:sequence>
</bpel:while>
<bpel:assign validate="no" name="Assign">
 <bpel:copy>
  <bpel:from><![CDATA[concat('Hello', $output.payload/tns:result, '! How are you?')]]></bpel:from>
  <bpel:to part="payload" variable="output">
   <bpel:query queryLanguage="urn:oasis:names:tc:wsbpel:2.0:sublang:xpath1.0"><![CDATA[tns:result]]></bpel:query>
  </bpel:to>
 </bpel:copy>
</bpel:assign>M

After deployment, for the above SOAP input all names were returned.

Creating arrays

OK. So here comes a really difficult part. Iterating over an array is a piece of cake. Creating arrays is much harder. Why? Because WS-BPEL spec does not say anything about instantiating arrays.

There are two things that can be done in order to create arrays:

use XSLT transformations
use vendor-specific extensions and XSLT functions

I will show you both.

Before I started creating arrays I added two new elements to my response for storing odd and even input words:

<element name="oddInput" maxOccurs="unbounded" type="string"/>
<element name="evenInput" maxOccurs="unbounded" type="string"/>

Manipulating arrays using XSLT transformations

For this task I used WS-BPEL function called bpel:doXslTransform. As a first argument in takes an XSTL stylesheet file, the second on is the XML input, the following arguments are XSLT params in name-value pairs.

So first, the XSLT:

<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" version="1.0">
 <xsl:output method="xml"/>
 <xsl:template match="/">
  <!-- <root /> element prevents org.w3c.dom.DOMException: HIERARCHY_REQUEST_ERR -->
  <root><xsl:apply-templates /></root>
 </xsl:template>
 <xsl:template
  match="*[local-name() = 'ArrayOperationsBusinessProcessRequest' and namespace-uri() = 'http://jee-bpel-soa.blogspot.com/ws-bpel/array-operations']">
   <xsl:apply-templates select="*[local-name() = 'input' and namespace-uri() = 'http://jee-bpel-soa.blogspot.com/ws-bpel/array-operations' and position() mod 2 = 0]" />
 </xsl:template>

 <xsl:template match="@*|node()">
  <xsl:copy>
   <xsl:apply-templates select="@*|node()"/>
  </xsl:copy>
 </xsl:template>
</xsl:stylesheet>

And the WS-BPEL invoking the XSLT:

<bpel:assign validate="no" name="Do XSLT Transform">
 <bpel:copy>
  <bpel:from>bpel:doXslTransform("ArrayCopy.xslt", $input.payload)
  </bpel:from>
  <bpel:to part="payload" variable="output">
   <bpel:query queryLanguage="urn:oasis:names:tc:wsbpel:2.0:sublang:xpath1.0">
                <![CDATA[tns:evenInput]]>
   </bpel:query>
  </bpel:to>
 </bpel:copy>
</bpel:assign>

No that bad. I mean, the XSLT is no that complicated, but it's not what you would like to do every time you need to create an array :)

Manipulating arrays using Apache ODE specific functions

OK. So Apache ODE provides ode namespace which contains many, many useful functions (I encourage you to check it out). Among those functions is function called: ode:insert-as-last-into. It expects two arguments, the first one is a destination array, the second one is an element to be inserted.

The usage of this function is straig forward. Using XSLT I copied even words, below I copy odd words:

<bpel:while name="While">
 <bpel:condition><![CDATA[$counter <= count($input.payload/tns:input)]]></bpel:condition>
 <bpel:sequence>
  <bpel:assign validate="no" name="Assign Output">
   <bpel:copy>
    <bpel:from>
                <![CDATA[concat($output.payload/tns:result, ' ', $input.payload/tns:input[number($counter)])]]>
    </bpel:from>
    <bpel:to part="payload" variable="output">
     <bpel:query queryLanguage="urn:oasis:names:tc:wsbpel:2.0:sublang:xpath1.0"><![CDATA[tns:result]]></bpel:query>
    </bpel:to>
   </bpel:copy>

  </bpel:assign>
  <bpel:if name="If">
   <bpel:condition><![CDATA[number($counter) mod 2 = 1]]></bpel:condition>
   <bpel:assign validate="no" name="Assign Odd">
    <bpel:copy>
     <bpel:from>
                        <![CDATA[ode:insert-as-last-into($output.payload/tns:oddInput, $input.payload/tns:input[number($counter)])]]>
     </bpel:from>
     <bpel:to part="payload" variable="output">
      <bpel:query queryLanguage="urn:oasis:names:tc:wsbpel:2.0:sublang:xpath1.0">
                            <![CDATA[tns:oddInput]]>
      </bpel:query>
     </bpel:to>
    </bpel:copy>
   </bpel:assign>
  </bpel:if>
 </bpel:sequence>
</bpel:while>

And that's all.

Summary

Using the soapUI, for the given input request:

<soapenv:Envelope xmlns:soapenv="http://schemas.xmlsoap.org/soap/envelope/" xmlns:arr="http://jee-bpel-soa.blogspot.com/ws-bpel/array-operations">
   <soapenv:Header/>
   <soapenv:Body>
      <arr:ArrayOperationsBusinessProcessRequest>
         <arr:input>Łukasz</arr:input>
         <arr:input>Jerzy</arr:input>
         <arr:input>Izydor</arr:input>
         <arr:input>Budnik</arr:input>
      </arr:ArrayOperationsBusinessProcessRequest>
   </soapenv:Body>
</soapenv:Envelope>

the WS-BPEL response was:

<soapenv:Envelope xmlns:soapenv="http://schemas.xmlsoap.org/soap/envelope/">
   <soapenv:Body>
      <ArrayOperationsBusinessProcessResponse xmlns="http://jee-bpel-soa.blogspot.com/ws-bpel/array-operations">
         <tns:result xmlns:tns="http://jee-bpel-soa.blogspot.com/ws-bpel/array-operations">Hello Łukasz Jerzy Izydor Budnik! How are you?</tns:result>
         <tns:oddInput xmlns:tns="http://jee-bpel-soa.blogspot.com/ws-bpel/array-operations">
            <input>Łukasz</input>
            <input>Izydor</input>
         </tns:oddInput>
         <tns:evenInput xmlns:tns="http://jee-bpel-soa.blogspot.com/ws-bpel/array-operations">
            <input xmlns:arr="http://jee-bpel-soa.blogspot.com/ws-bpel/array-operations">Jerzy</input>
            <input xmlns:arr="http://jee-bpel-soa.blogspot.com/ws-bpel/array-operations">Budnik</input>
         </tns:evenInput>
      </ArrayOperationsBusinessProcessResponse>
   </soapenv:Body>
</soapenv:Envelope>

So it worked!

Cheers!
Łukasz

Automatically starting up Java EE 5 Timers

2011-04-04T10:23:00.003+02:00

When you're developing a Java EE application and need a simple scheduler you may decide to use Java EE Timer Service API.

Of course Java EE Timer Service is not as powerful as Quartz, but it's provided out of the box :)

Last week I needed to implement a simple scheduler in my Java EE 5 app. I encountered a small problem - I wanted to start it automatically just after the application start up.

Here is how I did it.

1st approach: create timer inside @PostConstruct method

I tried this at first:

@Local
public interface PeriodicWorker {
 void callback(Timer timer);
}

@Stateless
public class PeriodicWorkerImpl implements PeriodicWorker {
 private static final Log log = LogFactory.getLog(PeriodicWorkerImpl.class);
 @Resource
 private TimerService timerService;
 @PostConstruct
 void init() {
  log.debug("Timer Service injected " + (timerService != null));
  timerService.createTimer(new Date(), 10000, PeriodicWorkerImpl.class);
 }
 @Timeout
 @Override
 public void callback(Timer timer) {
  log.debug("Got timeout from timer " + timer.getInfo());  
 }
}

Of course it didn't work. Java EE container does not instantiate all SLSB when the app is started those components are started when they are needed (lazy loading).

So my problem was, where to inject it...

2nd approach: injecting worker SLSB it into another EJB component

Injecting into another EJB? Might be tempting, but again, that EJB component had to be first called in order to be instantiated and all its @Resource fields populated.

But then it struck me, the reference to my SLSB can be injected into web container-managed ServletContextListener object.

3rd approach: injecting worker SLSB into ServletContextListener

ServletContextListener defines lifecycle methods which are called after the servlet context is initialised/destroyed.

I was almost done, but this time I got an exception saying that it is illegal to create a timer within SLSB's @PostConstruct method.

This time it was simple to fix.

4th and final solution

I added createTimer method to my PeriodicWorker and moved creation of timer there:

@Stateless
public class PeriodicWorkerImpl implements PeriodicWorker {
 private static final Log log = LogFactory.getLog(PeriodicWorkerImpl.class);
 @Resource
 private TimerService timerService;
 @PostConstruct
 void init() {
  log.debug("Timer Service injected " + (timerService != null));
 }
 @Override
 public void createTimer() {
  timerService.createTimer(new Date(), 10000, PeriodicWorkerImpl.class);
 }
 @Timeout
 @Override
 public void callback(Timer timer) {
  log.debug("Got timeout from timer " + timer.getInfo());  
 }
}

And then I simply call createTimer() method from my ServletContextListener.

public class TimerInitialiserContextListener implements ServletContextListener { 
 private static final Log log = LogFactory.getLog(TimerInitialiserContextListener.class);
 @EJB
 private PeriodicWorker worker;
 public void contextDestroyed(ServletContextEvent ctx) {
 }
 public void contextInitialized(ServletContextEvent ctx) {
  worker.createTimer();
 }
}

I deployed the app and it worked like a charm.

It turned out not to be that simple (took me 15 minutes including deploying on application server!), but it works out of the box. I pity there is no EJB-specific component which is started when the EJB module is started. Wouldn't it be nice?

Note: Yes it would, and it's possible in Java EE 6. See this post's comments below :)

cheers,
Łukasz

Manipulating WS-BPEL partnerLink

2011-03-21T07:50:00.001+01:00

Last week I was trying to copy partner link's endpoint reference to a variable. It's simple to override, but how to get it?

Section 3.4.5 of Web Services Business Process Execution Language Version 2.0 Primer states, that:

<from partnerLink="Supplier"/>

is sufficient.

Well, not in case of Apache ODE...

Running on Apache ODE

When you run the following code on Apache ODE:

<bpws:copy>
  <bpws:from partnerLink="abcPartnerLink"/>
  <bpws:to part="parameters" variable="variable">
   <bpws:query>namespace:EPR</bpws:query>
  </bpws:to>
</bpws:copy>

you will get:

java org.apache.ode.tools.bpelc.cline.BpelC: error: [CopyFromUndeclaredPartnerRole] Copy from partner link
abcPartnerLink with enpoint reference myRole requires the corresponding role to be defined on partner link.

Well, OK I understand the message, but how to fix it?

I tried google, but with no luck. Finally I found a solution for this. You have to add endpointReference attribute to <bpws:from /> element, just like this:

<bpws:from partnerLink="abcPartnerLink" endpointReference="partnerRole" />

Works like a charm.

Unexpected help

But guess who helped me with finding the answer to this question?

My "beloved" Eclipse BPEL designer. Remember my Dynamic Web Services invocations in WS-BPEL post where I cut Eclipse BPEL to the ribbons?

Well, there is a newer version of Eclipse BPEL designer and I must say that things improved greatly. During last month I created 3 complex WS-BPEL processes and I hardly had to modify anything by hand! I also liked the feature of asking whether I want to generate initializers for variables - cool. Good job Eclipse BPEL developers!

cheers,
Łukasz

Google Chrome is a piece of ...

2011-02-11T10:08:00.000+01:00

I switched from Firefox to Chrome some time ago, but there is one thing I really hate about Chrome. It's its complete infirmity when it comes to handling XML content.

To view and download WSDL files I have to use Internet Explorer. Why?

First, when you view WSDL in Chrome it displays an empty page, not XML tree, like for example FF or IE. And second, most terifying thing in handling XML content, Chrome is changing camel case elements' names to lowecase when you save/download XML documents. Here are examples of such elements:

<xs:complextype name="cancelBookingResponse">
    <xs:sequence>
      <xs:element name="result" type="xs:boolean" />
    </xs:sequence>
</xs:complextype>
 
<wsdl:porttype name="Warehouse">
     <wsdl:operation name="cancelBooking">
       <wsdl:input message="tns:cancelBooking" name="cancelBooking">
     </wsdl:input>
       <wsdl:output message="tns:cancelBookingResponse" name="cancelBookingResponse">
     </wsdl:output>
    </wsdl:operation>
</wsdl:porttype>

What a disgrace!

Long live IE! :)

Łukasz

Implementing simple Web Services Registry using CXF

2011-02-08T10:08:00.001+01:00

UDDI is an enterprise-grade solution for Web Services Directory. Most of Java EE application servers are shipped with UDDI applications. I wrote about UDDI and JAXR previously: Preparing for SCDJWS Part 8: UDDI, Fixing jUDDI 0.9rc4 and Scout 1.2 bugs, and Preparing for SCDJWS Part 15: JAXR and Web Services Registries.

Today I will show you a more flyweight approach to Web Services Registry. I will use Apache CXF for it.

Bean configuration

Let's start with beans.xml configuration:

<bean id="RegistryServiceBean" class="pl.gda.pg.eti.nuntius.testcases.orders_local_business_process_services.registry.RegistryServiceImpl " />
<jaxws:endpoint 
   id="RegistryService" 
   implementor="#RegistryServiceBean" 
   address="/Registry" />

:) Simple isn't it?

The implementation

Assume our registry service expects namespace and portType arguments, and returns a list of W3CEndpointReference (WS-Addressing) endpoints:

@WebService(name = "Registry", serviceName = "RegistryService")
public class RegistryServiceImpl implements ApplicationContextAware {

 private ServletTransportFactory servletTransportFactory;

 public List<W3CEndpointReference> findWebServices(
   @WebParam(name = "namespace") String namespace,
   @WebParam(name = "portType") String portType) {

  QName portTypeQName = new QName(namespace, portType);

  List<W3CEndpointReference> endpointReferences = new ArrayList<W3CEndpointReference>();

  for (ServletDestination servletDestination : servletTransportFactory
    .getDestinations()) {
   EndpointInfo endpointInfo = servletDestination.getEndpointInfo();
   if (portTypeQName.equals(endpointInfo.getInterface().getName())) {
    String address = endpointInfo.getAddress();
    W3CEndpointReference endpointReference = new W3CEndpointReferenceBuilder()
      .address(address).build();
    endpointReferences.add(endpointReference);
   }
  }

  return endpointReferences;
 }

 @Override
 @WebMethod(exclude = true)
 public void setApplicationContext(ApplicationContext applicationContext)
   throws BeansException {
  servletTransportFactory = (ServletTransportFactory) applicationContext
    .getBean(ServletTransportFactory.class.getCanonicalName());
 }

}

Summary

I think no comments are required, but if you have any questions give me a shout.

cheers,
Łukasz