Execute the slow integration tests separately from unit tests and show as much information about them as possible in Sonar.

A short definition of an Android Parcel would be that of a message container for lightweight, high-performance Inter-process communication (IPC). On Android, a "process" is a standard Linux one, and one process cannot normally access the memory of another process, so with Parcels, the Android system decomposes objects into primitives that can be marshaled/unmarshaled across process boundaries. But Parcels can also be used within the same process, to pass data across different components of a same application. As an example, a typical Android application has several screens, called "Activities" , and needs to communicate data or action from one Activity to the next. To write an object than can be passed through, we can implement the Parcelable interface. Android itself provides a built-in Parcelable object called an Intent which is used to pass information from one component to another. Using an Intent is pretty straightforward. Let's say we're collecting user data from our initial screen called CollectDataActivity. // inside CollectDataActivity, construct intent to pass along the next Activity, i.e. screen Intent in = new Intent(this, ProcessDataActivity.class); in.putExtra("userid", id); // (key,value) pairs in.putExtra("age", age); in.putExtra("phone", phone); in.putExtra("is_registered", true); // call next Activity --> next screen comes up startActivity(in); We need to collect that information from our data collection screen to process it. So all we do is the following: // inside ProcessDataActivity, get the info needed from previous Activity Intent in = this.getIntent(); in.getLongExtra("userid", 0L); in.getIntExtra("age", 0); in.getStringExtra("phone"); in.getBooleanExtra("is_registered", false); // false = default value overridden by user input Again, pretty straightforward. We retrieve the data using the same keys used to send it, and using our Intent's corresponding methods for each data type. But even when communicating with Intents, we can still use Parcels to pass data within the intent. For instance, we can do the above in a more elegant way using a custom, Parcelable User class: In the first Activity: // in CollectDataActivity, populate the Parcelable User object using its setter methods User usr = new User(); usr.setId(id); // collected from user input// etc.. // pass it to another component Intent in = new Intent(this, ProcessDataActivity.class); in.putExtra("user", usr); startActivity(in); In the second Activity: // in ProcessDataActivity retrieve User Intent intent = getIntent(); User usr = (User) intent.getParcelableExtra("user"); And this is what a Parcelable User class looks like: import android.os.Parcel; import android.os.Parcelable; public class User implements Parcelable { private long id; private int age; private String phone; private boolean registered; // No-arg Ctor public User(){} // all getters and setters go here //... /** Used to give additional hints on how to process the received parcel.*/ @Override public int describeContents() { // ignore for now return 0; } @Override public void writeToParcel(Parcel pc, int flags) { pc.writeLong(id); pc.writeInt(age); pc.writeString(phone); pc.writeInt( registered ? 1 :0 ); } /** Static field used to regenerate object, individually or as arrays */ public static final Parcelable.Creator CREATOR = new Parcelable.Creator() { public User createFromParcel(Parcel pc) { return new User(pc); } public User[] newArray(int size) { return new User[size]; } }; /**Ctor from Parcel, reads back fields IN THE ORDER they were written */ public User(Parcel pc){ id = pc.readLong(); age = pc.readInt(); phone = pc.readString(); registered = ( pc.readInt() == 1 ); } } What we did was: Make our User class implement the Parcelable interface. Parcelable is not a marker interface, hence what follows: Implement its describeContents method, which in this case does nothing. Implement its abstract method writeToParcel, which takes the current state of the object and writes it to a Parcel Add a static field called CREATOR to our class, which is an object implementing the Parcelable.Creator interface Add a Constructor that takes a Parcel as parameter. The CREATOR calls that constructor to rebuild our object. This looks like a lot of extra code at first, but bear in mind that, as in most cases, our application might evolve into incorporating more data from the user... Sometimes we need to pass complex objects from one component to another, and passing an object yields a cleaner design. The same logic applies for communicating between an Activity (foreground UI) and a background Service. We would just call the startService method instead of startActivity and pass it our Parcelable User object. Note that a Service is not running in a separate process by default. At this point, there are a couple of questions that may be raised: Isn't using an IPC-friendly, custom object for in-process communication simply overkill? Why would we want to use Parcelable, when we already have built-in Java serialization? The answer to the first concern is...maybe. But communicating through a custom object than through a list of key-value pairs is more OO, and it has no noticeable negative performance impact. As for the second question, why not simply have User implement Serializable, a theoretically simpler, marker interface? In one word, performance. Using Parcels is more efficient than serializing, at the price of some added complexity. That extra efficiency has in turn its limits: passing an image ( Bitmap) using Parcelable is generally not a good idea (although Bitmap does in fact implement Parcelable). A much more memory-efficient way would be to pass only its URI or Resource ID, so that other Android components in your application can have access to it. Another limitation of Parcelable is that it must not be used for general-purpose serialization to storage, since the underlying implementation may vary with different versions of the Android OS. So yes, Parcels are faster by design, but as high-performance transport, not as a replacement for general-purpose serialization mechanism. Having said all that, since our User object is Parcelable, it can now be sent from this application to another one running in another process, in particular through an interface implementing a remote service. In an upcoming post, we'll look at IPC and Android's Interface Definition Language (AIDL). from Tony's Blog

If you’re testing a web application, you can’t go far wrong with Selenium WebDriver. But in this web 2.0 world of ajax-y goodness, it can be a pain dealing with the asynchronous nature of modern sites. Back when all we had was web 1.0 you clicked a button and eventually you got a new page, or if you were unlucky: an error message. But now when you click links all sorts of funky things happen – some of which happen faster than others. From the user’s perspective this creates a great UI. But if you’re trying to automate testing this you can get all sorts of horrible race conditions. Thread.sleep The naive approach is to write your tests the same way you did before: you click buttons and assert that what you expected to happen actually happened. For the most part, this works. Sites are normally fast enough, even in a continuous integration environment, that by the time the test harness looks for a change it’s already happened. But then… things slow down a little and you start getting flickers - tests that sometimes pass and sometimes fail. So you add a little delay. Just 500 milliseconds should do it, while you wait for the server to respond and update the page. Then a month later it’s flickering again, so you make it 1 second. Then two… then twenty. The trouble is, each test runs at the pace that it runs at its slowest. If login normally takes 0.1 seconds, but sometimes takes 10 seconds when the environment’s overloaded – the test has to wait for 10 seconds so as not to flicker. This means even though the app often runs faster, the test has to wait just in case. Before you know it, your tests are crawling and take hours to run – you’ve lost your fast feedback loop and developers no longer trust the tests. An Example Thankfully WebDriver has a solution to this. It allows you to wait for some condition to pass, so you can use it to control the pace of your tests. To demonstrate this, I’ve created a simple web application with a login form – the source is available on github. The login takes a stupid amount of time, so the tests need to react to this so as not to introduce arbitrary waits. The application is very simple – a username and password field with an authenticate button that makes an ajax request to log the user in. If the login is successful, we update the screen to let the user know. The first thing is to write our test (obviously in the real world we’d have written the test before our production code, but its the test that’s interesting here not what we’re testing – so we’ll do it in the wrong order just this once): @Test public void authenticatesUser() { driver.get("http://localhost:8080/"); LoginPage loginPage = LoginPage.open(driver); loginPage.setUsername("admin"); loginPage.setPassword("password"); loginPage.clickAuthenticate(); Assert.assertEquals("Logged in as admin", loginPage.welcomeMessage()); } We have a page object that encapsulates the login functionality. We provide the username & password then click authenticate. Finally we check that the page has updated with the user message. But how have we dealt with the asynchronous nature of this application? WebDriverWait Through the magic of WebDriverWait we can wait for a function to return true before we continue: public void clickAuthenticate() { this.authenticateButton.click(); new WebDriverWait(driver, 30).until(accountPanelIsVisible()); } private Predicate accountPanelIsVisible() { return new Predicate() { @Override public boolean apply(WebDriver driver) { return isAccountPanelVisible(); } }; } private boolean isAccountPanelVisible() { return accountPanel.isDisplayed(); } Our clickAuthenticate method clicks the button then instructs WebDriver to wait for our condition to pass. The condition is defined via a predicate (c’mon Java where’s the closures?). The predicate is simply a method that will run to determine whether or not the condition is true yet. In this case, we delegate to the isAccountPanelVisible method on the page object. This does exactly what it says on the tin, it uses the page element to check whether it’s visible yet. Simple, no? In this way we can define a condition we want to be true before we continue. In this case, the exit condition of the clickAuthenticate method is that the asynchronous authentication process has completed. This means that tests don’t need to worry about the internal mechanics of the page – about whether the operation is asynchronous or not. The test merely specifies what to test, the page object encapsulates how to do it. Javascript It’s all well and good waiting for elements to be visible or certain text to be present, but sometimes we might want more subtle control. A good approach is to update Javascript state when an action has finished. This means that tests can inspect javascript variables to determine whether something has completed or not – allowing very clear and simple coordination between production code and test. Continuing with our login example, instead of relying on a becoming visible, we could instead have set a Javascript variable. The code in fact does both, so we can have two tests. The second looks as follows: public void authenticate() { this.authenticateButton.click(); new WebDriverWait(driver, 30).until(authenticated()); } private Predicate authenticated() { return new Predicate() { @Override public boolean apply(WebDriver driver) { return isAuthenticated(); } }; } private boolean isAuthenticated() { return (Boolean) executor().executeScript("return authenticated;"); } private JavascriptExecutor executor() { return (JavascriptExecutor) driver; } This example follows the same basic pattern as the test before, but we use a different predicate. Instead of checking whether an element is visible or not, we instead get the status of a Javascript variable. We can do this because each WebDriver also implements the JavascriptExecutor allowing us to run Javascript inside the browser within the context of the test. I.e. the script “return authenticated” runs within the browser, but the result is returned to our test. We simply inspect the state of a variable, which is false initially and set to true once the authentication process has finished. This allows us to closely coordinate our production and test code without the risk of flickering tests because of race conditions. From http://blog.activelylazy.co.uk/2012/01/29/testing-asynchronous-applications-with-webdriverwait/

Tests developed during TDD can be classified into several levels, depending on the size of the object graph they need to work with. End-to-end tests span the whole application graph, while unit tests usually target a single public class at a time. In the middle we find functional tests, which exercise a group of objects. A recurring problem is that of nearby classes C creeping into unit tests of unrelated classes; this situation transform what would be a unit test of the original class O into a functional tests of O and C together (possibly with multiple C classes involved). Functional tests are handy for specifying behavior at an higher level of abstraction than that of a single object, and sometimes for checking the wiring of a component of the application. However, if they are introduced involuntarily in place of unit tests they are prone to raise maintenance problems, since they will need to change every time the C class is updated. Moreover, they will fail along with the unit test of C, pointing to a problem into either O or C, which are not able to localize immediately. Consider this test, where the original class is DocumentsDeclarationNodeCommand and the collaborating one is InMemoryDocumentCopy: @Test public void shouldSendTheListOfDocumentsAndWaitForAcknowledgement() throws ConnectionClosedException { UpstreamConnection upstream = mock(UpstreamConnection.class); DownstreamConnection downstream = mock(DownstreamConnection.class); InMemoryDocumentCopy first = new InMemoryDocumentCopy("1.txt", "hello"); InMemoryDocumentCopy second = new InMemoryDocumentCopy("2.txt", "hello2"); DocumentsDeclarationNodeCommand command = DocumentsDeclarationNodeCommand.fromDocumentCopies( Arrays.asList(first, second), 10001); command.execute(upstream, downstream); InOrder inOrder = inOrder(upstream, downstream); inOrder.verify(upstream).command("DOCUMENTS|PORT=10001"); inOrder.verify(upstream).command("1.txt|5"); inOrder.verify(upstream).command("2.txt|6"); inOrder.verify(upstream).endCommandSection(); inOrder.verify(downstream, times(1)).readResponse(); } The two expectations on command() make this test a functional one: a change in the textual serialization format of InMemoryDocumentCopy (such as "1.txt|sha1_hash|5") will break this checks, even if DocumentsDeclarationNodeCommand still works. Yet we cannot avoid to verify that the documents are really sent to the server by this object. Functional tests can be transformed again into unit tests by testing O with a Test Double instead of C (a Stub, or a Mock.) The only remaining dependency will be the one of the interface of C, which can be even extracted into an independent entity (a first-class interface in language that support them such as Java, C# and PHP.) Pure functions What happens when you can't easily inject a Test Double to maintain the tests at the unit level? This issues exists in functional languages where functions call a tree of other functions. A analogue approach to dependency injection is to inject the function as a parameter, but doesn't probably scale to the level of injection we perform on objects: every function signature would have to receive all the collaborating ones as additional parameters. There are even mocking frameworks for functional languages like Marick's one which are able to isolate a function from its collaborators. Uncle Bob uses the Derived Expectation pattern instead: testing "update-all" (let [ o1 (make-object ...) o2 (make-object ...) o3 (make-object ...) os [o1 o2 o3] us (update-all os) ] (is (= (nth us 0) (reposition (accelerate (accumulate-forces os o1) (is (= (nth us 1) (reposition (accelerate (accumulate-forces os o2) (is (= (nth us 2) (reposition (accelerate (accumulate-forces os o3) ) ) The update-all function calls internally reposition, accelerate and accumulate-forces (or it calls other functions which in turn call these three). Instead of specifying unreadable literal expectations in the tests like (1.096, 4.128), this approach let the test specify update-all link to the other functions without introducing magic numbers. It is therefore a unit test for update-all, while the same test containing numbers would be a functional test. Note that this approach is safe for functional languages because the collaborating functions have no state, being pure; you can call reposition and accelerate how many times you want, and their result won't change. This is not necessarily true for collaborators in object-oriented languages: in principle, a method can return a different value for each call. Tests with derived expectations As long as the composed methods do not change their result, this approach would build real unit tests, whose success does not depend on the correctness of classes other than the one under test. Apart from corner cases like the composed methods throwing exceptions, a change in the collaborator's behavior would change only the collaborator's test. Value Objects are the ideal collaborator to stub out with derived expectations: they are immutable, so their methods always return the same result. Their code is usually self-contained and simple, so it's difficult for a method to throw an exception or to break internally once the Value Object has been correctly built. Being simple, final classes they do not implement an explicit interface; and they are not commonly substituted by Test Doubles. Their behavior is mixed in with the objects using them. The test becomes: @Test public void shouldSendTheListOfDocumentsAndWaitForAcknowledgement() throws ConnectionClosedException { UpstreamConnection upstream = mock(UpstreamConnection.class); DownstreamConnection downstream = mock(DownstreamConnection.class); InMemoryDocumentCopy first = new InMemoryDocumentCopy("1.txt", "hello"); InMemoryDocumentCopy second = new InMemoryDocumentCopy("2.txt", "hello2"); DocumentsDeclarationNodeCommand command = DocumentsDeclarationNodeCommand.fromDocumentCopies( Arrays.asList(first, second), 10001); command.execute(upstream, downstream); InOrder inOrder = inOrder(upstream, downstream); inOrder.verify(upstream).command("DOCUMENTS|PORT=10001"); inOrder.verify(upstream).command(first.toString()); inOrder.verify(upstream).command(second.toString()); inOrder.verify(upstream).endCommandSection(); inOrder.verify(downstream, times(1)).readResponse(); } Conclusion We saw that Test Doubles like Mocks and Stubs are not the only way to achieve isolated tests, which fail only where the class under test fail and not when a collaborator changes its implementation. In the Example, DocumentsDeclarationNodeCommand is tested by involving the real collaborator, but setting up Derived Expectation from it instead of literal ones. The result is this test is only tied to the method signatures of the collaborator instead of to the real behavior (the output format of toString()). This technique doesn't need to be used often: its purpose is to isolate from an immutable object, without introducing a Test Double.

To test mobile versions of websites, it is useful to be able to connect to a web server on your local machine from a web browser on an Android emulator without having to expose the web server to the Internet. You can’t use the normal loop-back IP address of 127.0.0.1 because that refers to the emulated Android device itself. Instead you have to use 10.0.2.2 to connect to the host machine. That’s fine if your local web server is serving a single site, but if you are using name-based virtual hosting to serve different sites depending on the host name of the request (with aliases for localhost defined in your machine’s hosts file), then you need to be making requests from the browser using the correct host name, not the IP address. The Android emulator does not use the host machine’s hosts file for name resolution so attempting to access http://myvirtualhost in the emulator’s browser will not work. This is because the emulated Android device has it’s own hosts file, so you have to update this to map the virtual host names to the local machine. The first step is to start the AVD with an increased partition size otherwise you may get an out of memory error when you try to save the modified hosts file: emulator -avd MyAVD -partition-size 128 You then have to remount the system partition so that it is writeable: adb remount Then copy the hosts file from the emulated device to the host machine: adb pull /etc/hosts Edit the hosts file so that it includes mappings for all relevant virtual host names: 127.0.0.1 localhost 10.0.2.2 myvirtualhost1 myvirtualhost2 Then copy the updated file back to the emulated device: adb push hosts /etc/hosts You should then be able to visit http://myvirtualhost1 in the emulator’s browser and see the correct site. From http://blog.uncommons.org/2012/01/12/accessing-local-name-based-virtual-hosts-from-the-android-emulator/

clover is a great tool for generating code coverage reports from your unit tests. it can be executed as a plugin in eclipse, maven or ant. however, not everyone knows that it can also be used to collect coverage data of integration tests. this post explains how to collect coverage data with clover at runtime. this post assumes that you already know what are unit and integration tests. this post assumes that you know what clover is, and already used it either with eclipse, ant or maven. * let me assured you that even though the directions bellow seems complicated and clumsy at first, after doing them once or twice it is really easy to repeat them. motivation the default action of clover is to gather code coverage information during build time or compile time. therefore, this information includes just the coverage data created by unit tests. if you are developing web applications, you probably use more technologies to test your applications beside unit tests. these technologies may include httpunit/ htmlunit or automation technologies (like selenium). these technologies do not work at build time, they can only work during run time, where a web server is up and running and http calls are made. as a result, the code coverage made during build time is not reflecting the actual code coverage. we should be able to test the coverage while a server is running. the idea the idea is that we will first run clover regularly during build time. we will than take the clover artifacts, put them in our server and then run the integration tests. while running the integration tests, the clover database will be updated and we would be able to generate reports from it which will reflect both unit and integration tests. step 1 – preparation make sure that you have a web/application server (tomcat/jboss/weblogic…) with your web application already deployed. execute clover on your application as you would normally do (either by compiling the code on eclipse or by building with maven or ant, it doesn’t matter). the result of this action would be: clover db files. one of the outcome of executing the clover on your code are the db files. the db files hold all the information about your code and the coverage itself. the location of those files may change depending on the way you use clover and according to the way you configured clover in your environment. this is how the files looks like the .db file holds the information regarding your code (classes, methods and so on). all the other files hold all the coverage data. it is important that you will locate those files because we are going to use them in the next step. an instrumented code. another outcome of clover is that it instruments your code. a clover call is injected into each method so it would be reported in the coverage calculation. we will need this code. we will use this instrumented code in runtime to update the coverage data. if you use eclipse than the generated classes would be instrumented. if you use maven or ant than most chances are that a jar with all the instrumented code would be generated separately. search the instrument code jar. again, i can’t tell you exactly where it is located, but usually it generates a jar with a ‘clover’ postfix. example: if your jar name is my_app.jar, than the generated instrumented code jar will probably be something like my_app-clover.jar. so you will need to do some detective work here to find the instrumented classes/jar. if you are not sure whether the classes are instrumented or not, just decompile one of them with jad and search for the word clover inside of it. a code coverage report. this is a report with the unit test coverage. we don’t really need this, but it would be good so that we would be able to compare it with the report we will generate at the end. step 2 – updating the server the next steps are very important, please make sure you do them properly. replacing the existing application jar/classes with the instrumented jar/classes. take the instrumented jar/classes that were created by the clover and add it to your server’s classpath instead of the original jar/classes. the instrumented code will cause the clover db to be updated with the runtime data. adding clover jars and license. since we will use clover on runtime we will need also the clover jars in our server’s classpath. so add the clover own jars to your server’s classpath. if you are using clover with eclipse than these jars are located in the plugin folder of eclipse. if you are using maven than they will be loacted in your repository. * make sure you are using the same version of clover in your server as you used to generate the db files and instrumented code, otherwise it will not work and you will get error messages. also add the clover license file to the same location as the jars. adding the clover java argument. add the following java arg -dclover.initstring.basedir={ location of the db files that were created by the clover } . * notice – the path you have entered above is the path of the folder which contains the db files. example: -dclover.initstring.basedir=c:/workspace/mywebapp/target/clover. this java argument is used by the clover to locate the db files and update them. step 3 – restarting the server and running the tests now that hopefully all is set properly all that you need to do now is to restart your server and than running your integration tests. the tests should trigger the instrumented code which will call the clover api’s and will update the clover db. while running the tests: look at your log/console and search for error messages from clover. look at the folder which holds the clover db files. if everything is going as it should, new files will be created in this folder while running the tests. if not everything is going well the first time, don’t discourage, just go over each of the steps again. step 4 – generating an updated report if everything went well and new files were created in the db folder than that means you just need to generate a new report. if you are using clover with eclipse than you can simply push the reload button to reload the coverage data. if you are using maven or ant you can execute just the task which generates the report. another way is to use the clover htmlreporter to generate a report easily. now compare the new report to the old report. you should see that the new report coverage is much bigger than the old one since it contains also the integration tests coverage. * notice that not all the data is updated, even though the percentages are being updated, for some reason the calls counter does not. to summarize. as mentioned; yes, these instructions seems a bit complicated but after you succeed the first time, it is very easy to repeat it. in the company i work for we even made this whole process automatic and we are able to generate a full coverage report with unit and integrated tests combined. source: http://www.aviyehuda.com/2011/12/use-clover-to-generate-code-coverage-reports-of-your-integrationautomation-tests/

Introduction A common thing I come across is that teams using a mocking framework assume they are mocking. They are not aware that Mocks are just one of a number of 'Test Doubles' which Gerard Meszaros has categorised at xunitpatterns.com. It’s important to realise that each type of test double has a different role to play in testing. In the same way that you need to learn different patterns or refactoring’s, you need to understand the primitive roles of each type of test double. These can then be combined to achieve your testing needs. I'll cover a very brief history of how this classification came about, and how each of the types differs. I'll do this using some short, simple examples in Mockito. A Very Brief History For years people have been writing lightweight versions of system components to help with testing. In general it was called stubbing. In 2000' the article 'Endo-Testing: Unit Testing with Mock Objects' introduced the concept of a Mock Object. Since then Stubs, Mocks and a number of other types of test objects have been classified by Meszaros as Test Doubles. This terminology has been referenced by Martin Fowler in "Mocks Aren't Stubs" and is being adopted within the Microsoft community as shown in "Exploring The Continuum of Test Doubles" A link to each of these important papers are shown in the reference section. Categories of test doubles The diagram above shows the commonly used types of test double. The following URL gives a good cross reference to each of the patterns and their features as well as alternative terminology. http://xunitpatterns.com/Test%20Double.html Mockito Mockito is a test spy framework and it is very simple to learn. Notable with Mockito is that expectations of any mock objects are not defined before the test as they sometimes are in other mocking frameworks. This leads to a more natural style(IMHO) when beginning mocking. The following examples are here purely to give a simple demonstration of using Mockito to implement the different types of test doubles. There are a much larger number of specific examples of how to use Mockito on the website. http://docs.mockito.googlecode.com/hg/latest/org/mockito/Mockito.html Test Doubles with Mockito Below are some basic examples using Mockito to show the role of each test double as defined by Meszaros. I’ve included a link to the main definition for each so you can get more examples and a complete definition. Dummy Object http://xunitpatterns.com/Dummy%20Object.html This is the simplest of all of the test doubles. This is an object that has no implementation which is used purely to populate arguments of method calls which are irrelevant to your test. For example, the code below uses a lot of code to create the customer which is not important to the test. The test couldn't care less which customer is added, as long as the customer count comes back as one. public Customer createDummyCustomer() { County county = new County("Essex"); City city = new City("Romford", county); Address address = new Address("1234 Bank Street", city); Customer customer = new Customer("john", "dobie", address); return customer; } @Test public void addCustomerTest() { Customer dummy = createDummyCustomer(); AddressBook addressBook = new AddressBook(); addressBook.addCustomer(dummy); assertEquals(1, addressBook.getNumberOfCustomers()); } We actually don't care about the contents of customer object - but it is required. We can try a null value, but if the code is correct you would expect some kind of exception to be thrown. @Test(expected=Exception.class) public void addNullCustomerTest() { Customer dummy = null; AddressBook addressBook = new AddressBook(); addressBook.addCustomer(dummy); } To avoid this we can use a simple Mockito dummy to get the desired behaviour. @Test public void addCustomerWithDummyTest() { Customer dummy = mock(Customer.class); AddressBook addressBook = new AddressBook(); addressBook.addCustomer(dummy); Assert.assertEquals(1, addressBook.getNumberOfCustomers()); } It is this simple code which creates a dummy object to be passed into the call. Customer dummy = mock(Customer.class); Don't be fooled by the mock syntax - the role being played here is that of a dummy, not a mock. It's the role of the test double that sets it apart, not the syntax used to create one. This class works as a simple substitute for the customer class and makes the test very easy to read. Test stub http://xunitpatterns.com/Test%20Stub.html The role of the test stub is to return controlled values to the object being tested. These are described as indirect inputs to the test. Hopefully an example will clarify what this means. Take the following code public class SimplePricingService implements PricingService { PricingRepository repository; public SimplePricingService(PricingRepository pricingRepository) { this.repository = pricingRepository; } @Override public Price priceTrade(Trade trade) { return repository.getPriceForTrade(trade); } @Override public Price getTotalPriceForTrades(Collection trades) { Price totalPrice = new Price(); for (Trade trade : trades) { Price tradePrice = repository.getPriceForTrade(trade); totalPrice = totalPrice.add(tradePrice); } return totalPrice; } The SimplePricingService has one collaborating object which is the trade repository. The trade repository provides trade prices to the pricing service through the getPriceForTrade method. For us to test the business logic in the SimplePricingService, we need to control these indirect inputs i.e. inputs we never passed into the test. This is shown below. In the following example we stub the PricingRepository to return known values which can be used to test the business logic of the SimpleTradeService. @Test public void testGetHighestPricedTrade() throws Exception { Price price1 = new Price(10); Price price2 = new Price(15); Price price3 = new Price(25); PricingRepository pricingRepository = mock(PricingRepository.class); when(pricingRepository.getPriceForTrade(any(Trade.class))) .thenReturn(price1, price2, price3); PricingService service = new SimplePricingService(pricingRepository); Price highestPrice = service.getHighestPricedTrade(getTrades()); assertEquals(price3.getAmount(), highestPrice.getAmount()); } Saboteur Example There are 2 common variants of Test Stubs: Responder’s and Saboteur's. Responder's are used to test the happy path as in the previous example. A saboteur is used to test exceptional behaviour as below. @Test(expected=TradeNotFoundException.class) public void testInvalidTrade() throws Exception { Trade trade = new FixtureHelper().getTrade(); TradeRepository tradeRepository = mock(TradeRepository.class); when(tradeRepository.getTradeById(anyLong())) .thenThrow(new TradeNotFoundException()); TradingService tradingService = new SimpleTradingService(tradeRepository); tradingService.getTradeById(trade.getId()); } Mock Object http://xunitpatterns.com/Mock%20Object.html Mock objects are used to verify object behaviour during a test. By object behaviour I mean we check that the correct methods and paths are excercised on the object when the test is run. This is very different to the supporting role of a stub which is used to provide results to whatever you are testing. In a stub we use the pattern of defining a return value for a method. when(customer.getSurname()).thenReturn(surname); In a mock we check the behaviour of the object using the following form. verify(listMock).add(s); Here is a simple example where we want to test that a new trade is audited correctly. Here is the main code. public class SimpleTradingService implements TradingService{ TradeRepository tradeRepository; AuditService auditService; public SimpleTradingService(TradeRepository tradeRepository, AuditService auditService) { this.tradeRepository = tradeRepository; this.auditService = auditService; } public Long createTrade(Trade trade) throws CreateTradeException { Long id = tradeRepository.createTrade(trade); auditService.logNewTrade(trade); return id; } The test below creates a stub for the trade repository and mock for the AuditService We then call verify on the mocked AuditService to make sure that the TradeService calls it's logNewTrade method correctly @Mock TradeRepository tradeRepository; @Mock AuditService auditService; @Test public void testAuditLogEntryMadeForNewTrade() throws Exception { Trade trade = new Trade("Ref 1", "Description 1"); when(tradeRepository.createTrade(trade)).thenReturn(anyLong()); TradingService tradingService = new SimpleTradingService(tradeRepository, auditService); tradingService.createTrade(trade); verify(auditService).logNewTrade(trade); } The following line does the checking on the mocked AuditService. verify(auditService).logNewTrade(trade); This test allows us to show that the audit service behaves correctly when creating a trade. Test Spy http://xunitpatterns.com/Test%20Spy.html It's worth having a look at the above link for the strict definition of a Test Spy. However in Mockito I like to use it to allow you to wrap a real object and then verify or modify it's behaviour to support your testing. Here is an example were we check the standard behaviour of a List. Note that we can both verify that the add method is called and also assert that the item was added to the list. @Spy List listSpy = new ArrayList(); @Test public void testSpyReturnsRealValues() throws Exception { String s = "dobie"; listSpy.add(new String(s)); verify(listSpy).add(s); assertEquals(1, listSpy.size()); } Compare this with using a mock object where only the method call can be validated. Because we only mock the behaviour of the list, it does not record that the item has been added and returns the default value of zero when we call the size() method. @Mock List listMock = new ArrayList(); @Test public void testMockReturnsZero() throws Exception { String s = "dobie"; listMock.add(new String(s)); verify(listMock).add(s); assertEquals(0, listMock.size()); } Another useful feature of the testSpy is the ability to stub return calls. When this is done the object will behave as normal until the stubbed method is called. In this example we stub the get method to always throw a RuntimeException. The rest of the behaviour remains the same. @Test(expected=RuntimeException.class) public void testSpyReturnsStubbedValues() throws Exception { listSpy.add(new String("dobie")); assertEquals(1, listSpy.size()); when(listSpy.get(anyInt())).thenThrow(new RuntimeException()); listSpy.get(0); } In this example we again keep the core behaviour but change the size() method to return 1 initially and 5 for all subsequent calls. public void testSpyReturnsStubbedValues2() throws Exception { int size = 5; when(listSpy.size()).thenReturn(1, size); int mockedListSize = listSpy.size(); assertEquals(1, mockedListSize); mockedListSize = listSpy.size(); assertEquals(5, mockedListSize); mockedListSize = listSpy.size(); assertEquals(5, mockedListSize); } This is pretty Magic! Fake Object http://xunitpatterns.com/Fake%20Object.html Fake objects are usually hand crafted or light weight objects only used for testing and not suitable for production. A good example would be an in-memory database or fake service layer. They tend to provide much more functionality than standard test doubles and as such are probably not usually candidates for implementation using Mockito. That’s not to say that they couldn’t be constructed as such, just that its probably not worth implementing this way. References Test Double Patterns Endo-Testing: Unit Testing with Mock Objects Mock Roles, Not Objects Mocks Aren't Stubs http://msdn.microsoft.com/en-us/magazine/cc163358.aspx Original Article. The original article and others can be found here http://johndobie.blogspot.com/2011/11/test-doubles-with-mockito.html

have been working on a fairly simple JMeter load script that I can run a series of 4 sequential pages against an Apache server, but the goal was to have the server support 2,000 concurrent requests for 5 minutes without error. Most of my issues in this exercise have been with JMeter and the client machine used to test the Apache server. To begin, I must state I was originally configuring Apache with a prefork MPM: StartServers 100 MinSpareServers 75 MaxSpareServers 100 ServerLimit 2000 MaxClients 2000 MaxRequestsPerChild 0 At approximately line 72 of Jmeter.bat, there are several entries the manage the JVM for running Jmeter. set HEAP=-Xms512m -Xmx512m set NEW=-XX:NewSize=128m -XX:MaxNewSize=128m set SURVIVOR=-XX:SurvivorRatio=8 -XX:TargetSurvivorRatio=50% set TENURING=-XX:MaxTenuringThreshold=2 set RMIGC=-Dsun.rmi.dgc.client.gcInterval=600000 -Dsun.rmi.dgc.server.gcInterval=600000 set PERM=-XX:PermSize=64m -XX:MaxPermSize=64m I decided to start with 1,000 concurrent requests for 5 minutes just to see how the test would fair. With the above settings I started getting OOM errors almost immediately so I decided to increase the HEAP and NEW memory to eliminate the issue and wanted to add more GC settings to increase the JVM’s ability to clean up: set HEAP=-Xms1024m -Xmx1024m -Xss128k set NEW=-XX:NewSize=256m -XX:MaxNewSize=256m set SURVIVOR=-XX:SurvivorRatio=14 -XX:TargetSurvivorRatio=50% set "TENURING=-XX:+UseConcMarkSweepGC -XX:+UseParNewGC -XX:+CMSParallelRemarkEnabled -XX:+UseCMSCompactAtFullCollection -XX:+DisableExplicitGC -XX:+UseCMSInitiatingOccupancyOnly -XX:CMSInitiatingOccupancyFraction=70 -XX:MaxTenuringThreshold=4" set "EVACUATION=-XX:+AggressiveOpts -XX:+UseFastAccessorMethods -XX:+UseCompressedStrings -XX:+OptimizeStringConcat" set RMIGC=-Dsun.rmi.dgc.client.gcInterval=600000 -Dsun.rmi.dgc.server.gcInterval=600000 set PERM=-XX:PermSize=64m -XX:MaxPermSize=64m This did resolve the JMeter OOM issues, but now started getting Apache errors. During the ramp-up phase, I started getting connection refused errors: Response code: Non HTTP response code: org.apache.http.conn.HttpHostConnectException Response message: Non HTTP response message: Connection to http://pasundtastgprt2:8001 refused I started looking at the Apache server and noticed that the number of httpd threads was at 1,000 and it appeared that JMeter was running out of memory because the requests where starting to back up. This is why we load test right! So I decided to run a worker MQM and recompiled Apache to support the new MPM ServerLimit 80 StartServers 25 MaxClients 2000 MinSpareThreads 75 MaxSpareThreads 125 ThreadsPerChild 5 MaxRequestsPerChild 0 I started testing this configuration and while monitoring the server running 1,000 concurrent requests and the server looked like Apache was handling 1,000 requests just fine. I was running a simple command to output the sockets and httpd processes on the server during the load test: while true do echo -----`date '+%r'` -----: netstat -ant | awk '{print $6}' | sort | uniq -c | sort -n echo httpd processes: [`ps aux | grep httpd | wc -l`] echo . sleep 30 done Then when I was monitoring the load test, I was concerned about seeing 82 httpd processes running which was the ServerLimit I had set. -----08:02:37 AM -----: 1 established) 1 Foreign 4 CLOSE_WAIT 17 LISTEN 32 FIN_WAIT2 41 ESTABLISHED 69 FIN_WAIT1 630 SYN_RECV 45386 TIME_WAIT [82] httpd processes . I now increased the load to my target of 2,000 concurrent requests and restarted the JMeter test and was able to get to around 1,800 concurrent request and started getting connection refused errors again. I suspected that my Servers where maxed out and was not able to create anymore threads for those servers where having: 80 server * 5 threads each server == 400 requests processed concurrently So I increased the number of threads to 25 80 server * 25 threads each server == 2,000 requests processed concurrently To end up with this worker setting: ServerLimit 80 StartServers 25 MaxClients 2000 MinSpareThreads 75 MaxSpareThreads 125 ThreadsPerChild 25 MaxRequestsPerChild 0 I was then able to turn the load up to 2,000 concurrent requests. -----08:53:27 AM -----: 1 established) 1 FIN_WAIT2 1 Foreign 4 CLOSE_WAIT 12 CLOSING 17 LISTEN 129 ESTABLISHED 621 FIN_WAIT1 1203 SYN_RECV 55556 TIME_WAIT [53] httpd processes At this point we are only using 53 Servers and 2,000 clients. The tests sustained zero errors for 5 minutes during the test. As a test I increased the ThreadsPerChild to 40 and run the load test against 3,000 concurrent requests. I was able to get to around 2,800 concurrent requests then I started getting connection refused errors: Error Count: 1 Response code: Non HTTP response code: org.apache.http.conn.HttpHostConnectException Response message: Non HTTP response message: Connection to http://pasundtastgprt2:8001 refused and I also started getting JMeter errors: Response code: Non HTTP response code: java.net.BindException Response message: Non HTTP response message: Address already in use: connect So in the furure I would like to see how much further I can push the server and I think I can get to 3,000 concurrent and most likely far more than that on my current installation. From http://www.baselogic.com/blog/development/adding-memory-jmeter/

This article describes some practices for test-driving multithreaded and distributed applications written in Java. The example I worked on and we will use is a peer-to-peer application composed of many Nodes (clients) and of a few Supernodes (servers). The ultimate goal it to build an application composed of all these entities, but the first tests target a Supernode serving one or more Nodes. The walking skeleton TDD is mostly iterative, but needs a starting point. The simplest story we can think of is that of a Node connecting to a Supernode. Client and servers usually run in their own threads (in the case of the server, multiple ones), but initially the Node object can just be a POJO and run in the test's thread because we do not need to manage multiple Nodes yet. The Supernode object instead is a Thread (or a Runnable) and so we already face a simplified version of the synchronization problem: how to make sure the Supernode is ready to answer to connections once we have started its thread? The JUnit test is the following: @Test public void aNodeCanConnectToASupernode() throws Exception { Supernode supernode = new Supernode(8888); supernode.start(); supernode.ensureStartupIsFinished(); Node n = new Node(); n.connect("127.0.0.1", 8888); assertEquals(1, supernode.getNodes()); } supernode.start() runs the new thread, while the call to ensureStartupIsFinished() will have to block until the other thread is ready. Then, we create a Node object and tell it to connect; after it has finished this operation, we count how many nodes have connected to the Supernode. To satisfy this test, the Supernode can be a single-threaded server: public class Supernode extends Thread { private int port; private boolean startupCompleted; private int nodes = 0; public Supernode(int port) { this.port = port; this.startupCompleted = false; } public void run() { ServerSocket sock; try { sock = new ServerSocket(this.port); // ...networking setup... synchronized (this) { this.startupCompleted = true; notify(); } while (true) { // ...accepting new connections on sock and other stuff } } public int getNodes() { return nodes; } synchronized public void ensureStartupIsFinished() throws InterruptedException { while (!this.startupCompleted) { wait(); } } } What's in this first example? Thread objects are manageable as POJOs from a single JVM: as long as we write them with this API it will be simple to instantiate and terminate them, and to add primitives for synchronization. The startupCompleted field, which is an example of this synchronization behavior added to the production code. Adding production code just for end-to-end testing purposes is not uncommon. The test thread blocks inside ensureStartupIsFinished() until it is woken up via notification. Even then, startupCompleted must be true or it will wait more. This is Plain Old Java Synchronization: note the synchronized blocks around this.wait() and this.notify(). The problem with frameworks and containers is you have to hope they provide the synchronization facilities to test your code once it's inside them: have you ever tried to wait for Tomcat to start? There are some noticeable missing parts in this code: the threads for each node. The current test does not require them as only one Node is connecting for now. Thread.sleep() calls: at least for the happy paths I have covered until now, I never need to introduce them and considered them a smell. Configuration files: if we had to read configuration, the tests would take really long to write and would refer continuously to external resources. This is the case when testing with external tools which are not embeddable (Tomcat requiring configuration files while Jetty allowing configuration to be passed in Java test code). You can always add file-based configuration later, but for now it will slow us down. Evolution By adding one test at the time with a larger scope, we can try to evolve the code and add the difficult networking, multithreading part one bit at a time. After some iterations, the test becomes: public class FileSharingNetworkTest { Supernode supernode; @Before public void setUp() throws Exception { supernode = new Supernode(8888); supernode.start(); supernode.ensureStartupIsFinished(); } @After public void tearDown() throws Exception { supernode.ensureStop(); } @Test public void aNodeCanConnectToASupernode() throws Exception { Node n = newNode(Arrays.asList("1.txt", "2.txt")); n.ensureConnectionIsFinished(); assertEquals(1, supernode.getNodes()); assertEquals(2, supernode.getDocuments()); } @Test public void multipleNodesCanConnectToASupernodeSimultaneously() throws Exception { Node n1 = newNode(); Node n2 = newNode(); n1.ensureConnectionIsFinished(); n2.ensureConnectionIsFinished(); assertEquals(2, supernode.getNodes()); } private Node newNode() { Node n = new Node("127.0.0.1", 8888); n.start(); n.setDocumentList(Arrays.asList("1.txt", "2.txt")); return n; } private Node newNode(List documentList) { Node n = new Node("127.0.0.1", 8888); n.start(); n.setDocumentList(documentList); return n; } } The server-side code doesn't have multiple threads yet. What is the test case that will call for them? You have to find it and write it. This workflow will ensure that there is a test that targets this case. In my case, it was the first test requiring interaction between the two clients, where one had to see the documents listed by the other after both had connected. Even if you know where you will end up, you can test-drive the implementation: the advantage is that you understand better a standard design and ensure its test coverage. After a few more tests, I have reached a multithreaded server with a main thread and chidren for managing the connections; and Node objects implemented as independent threads. Conclusions When working with TDD at a system scale that includes asynchronous behavior, we should strive for a test suite that is: fast; even with multiple threads to wait for, a single end to end test should take less than a second to complete. Comprehensive; TDD makes us only write tested code instead of copying down snippets from the web. Robust: totally deterministic, as every run will either pass or fail, even when repeated dozens of times. There should be no sleeping calls for all the happy paths; there should be synchronization and stopping facilities built into the system. Featuring unit tests: along with the end to end tests we should write unit tests for the objects we need to extract (and that will be single threaded). It was easy for me to get caught up into covering more and more cases with a full scale test, but unit tests are better at pointing out where a bug resides. We also have to keep in mind how to design our objects and interfaces: not starting with N threads but with at most 1 more than the test's one (the server or a remote peer). Evolving them: adding a few lines of verbose Java networking code each time. My example has evolved to N client threads, a server main thread and N server children threads talking with each client. I will now have to evolve it to a network of supernodes, being this about a file sharing network; to introduce secure channels and certificates. The difficult part is to constantly refactor to support new stories without having to rework the whole system for a single one. Not only extracting methods (an automated operation), but also to extract interfaces and most importantly objects; targeting the longest and complex classes and chopping them down into basic responsibilities.

This post investigates the performance of unmarshalling an XML document to Java objects using a number of different approaches. The XML document is very simple. It contains a collection of Person entities. person0 name0 person1 name1 ... There is a corresponding Person Java object for the Person entity in the XML ... @XmlAccessorType(XmlAccessType.FIELD) @XmlType(name = "", propOrder = { "id", "name" }) public class Person { private String id; private String name; public String getId() { return id; } public void setId(String id) { this.id = id; } public String getName() { return name; } public void setName(String value) { this.name = value; } } and a PersonList object to represent a collection of Persons. @XmlAccessorType(XmlAccessType.FIELD) @XmlRootElement(name = "persons") public class PersonList { @XmlElement(name="person") private List personList = new ArrayList(); public List getPersons() { return personList; } public void setPersons(List persons) { this.personList = persons; } } The approaches investigated were: Various flavours of JAXB SAX DOM In all cases, the objective was to get the entities in the XML document to the corresponding Java objects. The JAXB annotations on the Person and PersonList POJOS are used in the JAXB tests. The same classes can be used in SAX and DOM tests (the annotations will just be ignored). Initially the reference implementations for JAXB, SAX and DOM were used. The Woodstox STAX parsing was then used. This would have been called in some of the JAXB unmarshalling tests. The tests were carried out on my Dell Laptop, a Pentium Dual-Core CPU, 2.1 GHz running Windows 7. Test 1 - Using JAXB to unmarshall a Java File. @Test public void testUnMarshallUsingJAXB() throws Exception { JAXBContext jc = JAXBContext.newInstance(PersonList.class); Unmarshaller unmarshaller = jc.createUnmarshaller(); PersonList obj = (PersonList)unmarshaller.unmarshal(new File(filename)); } Test 1 illustrates how simple the progamming model for JAXB is. It is very easy to go from an XML file to Java objects. There is no need to get involved with the nitty gritty details of marshalling and parsing. Test 2 - Using JAXB to unmarshall a Streamsource Test 2 is similar Test 1, except this time a Streamsource object wraps around a File object. The Streamsource object gives a hint to the JAXB implementation to stream the file. @Test public void testUnMarshallUsingJAXBStreamSource() throws Exception { JAXBContext jc = JAXBContext.newInstance(PersonList.class); Unmarshaller unmarshaller = jc.createUnmarshaller(); StreamSource source = new StreamSource(new File(filename)); PersonList obj = (PersonList)unmarshaller.unmarshal(source); } Test 3 - Using JAXB to unmarshall a StAX XMLStreamReader Again similar to Test 1, except this time an XMLStreamReader instance wraps a FileReader instance which is unmarshalled by JAXB. @Test public void testUnMarshallingWithStAX() throws Exception { FileReader fr = new FileReader(filename); JAXBContext jc = JAXBContext.newInstance(PersonList.class); Unmarshaller unmarshaller = jc.createUnmarshaller(); XMLInputFactory xmlif = XMLInputFactory.newInstance(); XMLStreamReader xmler = xmlif.createXMLStreamReader(fr); PersonList obj = (PersonList)unmarshaller.unmarshal(xmler); } Test 4 - Just use DOM This test uses no JAXB and instead just uses the JAXP DOM approach. This means straight away more code is required than any JAXB approach. @Test public void testParsingWithDom() throws Exception { DocumentBuilderFactory domFactory = DocumentBuilderFactory.newInstance(); DocumentBuilder builder = domFactory.newDocumentBuilder(); Document doc = builder.parse(filename); List personsAsList = new ArrayList(); NodeList persons = doc.getElementsByTagName("person"); for (int i = 0; i persons = new ArrayList(); DefaultHandler handler = new DefaultHandler() { boolean bpersonId = false; boolean bpersonName = false; public void startElement(String uri, String localName,String qName, Attributes attributes) throws SAXException { if (qName.equalsIgnoreCase("id")) { bpersonId = true; Person person = new Person(); persons.add(person); } else if (qName.equalsIgnoreCase("name")) { bpersonName = true; } } public void endElement(String uri, String localName, String qName) throws SAXException { } public void characters(char ch[], int start, int length) throws SAXException { if (bpersonId) { String personID = new String(ch, start, length); bpersonId = false; Person person = persons.get(persons.size() - 1); person.setId(personID); } else if (bpersonName) { String name = new String(ch, start, length); bpersonName = false; Person person = persons.get(persons.size() - 1); person.setName(name); } } }; saxParser.parse(filename, handler); } The tests were run 5 times for 3 files which contain a collection of Person entities. The first first file contained 100 Person entities and was 5K in size. The second contained 10,000 entities and was 500K in size and the third contained 250,000 Person entities and was 15 Meg in size. In no cases was any XSD used, or any validations performed. The results are given in result tables where the times for the different runs are comma separated. TEST RESULTS The tests were first run using JDK 1.6.26, 32 bit and the reference implementation for SAX, DOM and JAXB shipped with JDK was used. Unmarshall Type 100 Persons time (ms) 10K Persons time (ms) 250K Persons time (ms) JAXB (Default) 48,13, 5,4,4 78, 52, 47,50,50 1522, 1457, 1353, 1308,1317 JAXB(Streamsource) 11, 6, 3,3,2 44, 44, 48,45,43 1191, 1364, 1144, 1142, 1136 JAXB (StAX) 18, 2,1,1,1 111, 136, 89,91,92 2693, 3058, 2495, 2472, 2481 DOM 16, 2, 2,2,2 89,50, 55,53,50 1992, 2198, 1845, 1776, 1773 SAX 4, 2, 1,1,1 29, 34, 23,26,26 704, 669, 605, 589,591 JDK 1.6.26 Test comments The first time unmarshalling happens is usually the longest. The memory usage for the JAXB and SAX is similar. It is about 2 Meg for the file with 10,000 persons and 36 - 38 Meg file with 250,000. DOM Memory usage is far higher. For the 10,000 persons file it is 6 Meg, for the 250,000 person file it is greater than 130 Meg. The performance times for pure SAX are better. Particularly, for very large files. The exact same tests were run again, using the same JDK (1.6.26) but this time the Woodstox implementation of StAX parsing was used. Unmarshall Type 100 Persons time (ms) 10K Persons time (ms) 250K Persons time (ms) JAXB (Default) 48,13, 5,4,4 78, 52, 47,50,50 1522, 1457, 1353, 1308,1317 JAXB(Streamsource) 11, 6, 3,3,2 44, 44, 48,45,43 1191, 1364, 1144, 1142, 1136 JAXB (StAX) 18, 2,1,1,1 111, 136, 89,91,92 2693, 3058, 2495, 2472, 2481 DOM 16, 2, 2,2,2 89,50, 55,53,50 1992, 2198, 1845, 1776, 1773 SAX 4, 2, 1,1,1 29, 34, 23,26,26 704, 669, 605, 589,591 JDK 1.6.26 + Woodstox test comments Again, the first time unmarshalling happens is usually proportionally longer. Again, memory usage for SAX and JAXB is very similar. Both are far better than DOM. The results are very similar to Test 1. The JAXB (StAX) approach time has improved considerably. This is due to the Woodstox implementation of StAX parsing being used. The performance times for pure SAX are still the best. Particularly for large files. The the exact same tests were run again, but this time I used JDK 1.7.02 and the Woodstox implementation of StAX parsing. Unmarshall Type 100 Persons time (ms) 10,000 Persons time (ms) 250,000 Persons time (ms) JAXB (Default) 165,5, 3, 3,5 611,23, 24, 46, 28 578, 539, 511, 511, 519 JAXB(Streamsource) 13,4, 3, 4, 3 43,24, 21, 26, 22 678, 520, 509, 504, 627 JAXB (StAX) 21,1,0, 0, 0 300,69, 20, 16, 16 637, 487, 422, 435, 458 DOM 22,2,2,2,2 420,25, 24, 23, 24 1304, 807, 867, 747, 1189 SAX 7,2,2,1,1 169,15, 15, 19, 14 366, 364, 363, 360, 358 JDK 7 + Woodstox test comments: The performance times for JDK 7 overall are much better. There are some anomolies - the first time the 100 persons and the 10,000 person file is parsed. The memory usage is slightly higher. For SAX and JAXB it is 2 - 4 Meg for the 10,000 persons file and 45 - 49 Meg for the 250,000 persons file. For DOM it is higher again. 5 - 7.5 Meg for the 10,000 person file and 136 - 143 Meg for the 250,000 persons file. Note: W.R.T. all tests No memory analysis was done for the 100 persons file. The memory usage was just too small and so it would have pointless information. The first time to initialise a JAXB context can take up to 0.5 seconds. This was not included in the test results as it only took this time the very first time. After that the JVM initialises context very quickly (consistly < 5ms). If you notice this behaviour with whatever JAXB implementation you are using, consider initialising at start up. These tests are a very simple XML file. In reality there would be more object types and more complex XML. However, these tests should still provide a guidance. Conclusions: The peformance times for pure SAX are slightly better than JAXB but only for very large files. Unless you are using very large files the performance differences are not worth worrying about. The progamming model advantages of JAXB win out over the complexitiy of the SAX programming model. Don't forget JAXB also provides random accses like DOM does. SAX does not provide this. Performance times look a lot better with Woodstox, if JAXB / StAX is being used. Performance times with 64 bit JDK 7 look a lot better. Memory usuage looks slightly higher. From http://dublintech.blogspot.com/2011/12/jaxb-sax-dom-performance.html

Neo4j is a high-performance, NOSQL graph database with all the features of a mature and robust database. In this post I will explain how to deploy a neo4j instance in Amazon EC2 web service. For this tutorial to take you no more than 10 minutes you should be able to execute properly some bash commands like mv, tar, ssh and scp (secure copy). I also assume that you have an account in Amazon Web Services and you are familiar to the process of launching instances. If not, I strongly recommend you to follow this starting guide and complete it till you manage to connect to your instance with ssh. Start downloading the latest stable version of neo4j. Which you can find here. The “Community Edition” fits well for development purposes. Do not forget to select the Unix version of the server. This will download a tar.gz file which you will copy to your EC2 instance later. While you download the neo4j server open the AWS Management Console and launch a Basic 32-bit Amazon Linux AMI. If you want to launch an Ubuntu AMI please notice that it doesn’t ship with Java, which is required for running neo4j. If you are not familiar with key pairs, pem files or security groups I insist you to follow the EC2 starting guide I mentioned above. You can either create a new security group or use the default, but you will need to configure a new security rule for the neo4j server port. After launching the instance, create a TCP rule on port 7474 with source 0.0.0.0/0. Here you are opening port 7474 for anyone. If you are planning to use the neo4j REST API and remotely call it from another server, for example a Rails application hosted in Heroku, for security reasons, you may want to change the source field to the address of your Heroku server. Do not forget to open port 22 (SSH), this is typically the first rule normal people create after launching an instance. You are almost done! You should now install neo4j in your instance. Open a terminal in your localhost and navigate to the path where you downloaded neo4j. Copy the file to your Amazon instance by using the scp command: scp -i your_pem_file.pem neo4j-community-1.6.M01-unix.tar.gz ec2-user@YOUR_PUBLIC_INSTANCE_DNS:/home/ec2-user Please notice that you will need to change the path to your pem file, typically placed in ~/.ssh, the filename of the neo4j server you just downloaded and the plublic DNS of your instance. Now connect to your instance with SSH: ssh -i your_pem_file.pem ec2-user@YOUR_PUBLIC_INSTANCE_DNS Untar the neo4j server: tar xvfz neo4j-community-1.6.M01-unix.tar.gz.tar.gz Move it to /usr/local and rename the folder to neo4j: sudo mv neo4j-community-1.6.M01 /usr/local/neo4j Almost done!!! You should now open neo4j-server.properties under the conf directory and add the following line: org.neo4j.server.webserver.address=0.0.0.0 This lines allows anyone to connect remotely to your neo4j database server. Now run the start script. From the neo4j server folder. sudo ./bin/neo4j start Finally, open a browser and access the webadmin interface of your neo4j database by typing http://YOUR_PUBLIC_INSTANCE_DNS:7474. You should see the Neo4j Monitoring and Management Tool, pretty cool! If not, ask me You can now try using the REST API and the curl bash command to insert nodes and relationships. I hope this post helped you, good luck! Follow me on Twitter @negarnil Source: http://www.cloudtmp.com/java/how-to-deploy-a-neo4j-instance-in-amazon-ec2-in-10-minutes/

So you're still interested in unit testing JavaScript (good). This post is an extension of my much more indepth first posting on how to unit test JavaScript using JS Test Driver. Please check it out here. Recap Last Posting In the last posting we successfully unit tested JavaScript using Maven and JsTest Driver. This allowes us to test JavaScript when on an environment that has a modern browser installed and can be run. Problem with typical CI environments So what happens when the test are passing on your local box, but you go to check in your code and the Continuous Integration (CI) server pukes on the new tests becasue there is no "screen" to run chrome or firefox? As of this posting, none of the top-tier browsers have a "headless" or an in-memory only browser window. There are alternatives to running JavaScript in a browser, such as rhino.js, env.js or HtmlUnit, however, these are just ports of browsers and the JavaScript and DOM representation are not 100% accurate which can lead to problems with your code when rendered in a client's browser. Approach What we need to do is to run JSTestDriver's browser in a Virtual X Framebuffer (Xvfb) which is possible on nearly all Linux based systems. The example below uses a Solaris version of Linux, however, Debian and RedHat linux distrubutions come with the simplified bash script to easily run an appliation in a virtual framebuffer. This solution was derived from one posted solution on the JS Test Driver wiki. The given example is also a full working example that is in use at my current client. Here is the quick list of what we will accomplish. Note, several of these steps are discussed in depth in the previous post and are not covered in depth here. Create a profile to run Js Unit-Tests Copy JsTestDriver library to a known location for Maven to use Copy JavaScript main and test files to known locations Use ANT to start JsTestDriver and pipe the screen into xvfb Here is a sample profile to use. You will need to adjust the properties at the top of the profile to match your system. ci-jstests /opt/swf/bin/firefox 1.3.2 /opt/X11R6/xvfb-run org.apache.maven.plugins maven-dependency-plugin 2.1 copy generate-resources copy com.google.jstestdriver jstestdriver ${js-test-driver.version} jar true jsTestDriver.jar ${project.build.directory}/jstestdriver false true maven-resources-plugin 2.4.3 copy-main-files generate-test-resources copy-resources ${project.build.directory}/test-classes/main-js src/main/webapp/scripts false copy-test-files generate-test-resources copy-resources ${project.build.directory}/test-classes/test-js src/test/webapp/scripts false org.apache.maven.plugins maven-antrun-plugin 1.6 test run Possible problems Although I cannot predict or fix all problems, I can share the one major problem I ran into with Solaris and the script used to fix that. In Solaris (and could happen to other distros) the xvfb-run script was not available and several of the other libraries did not exist. I first had to download the latest X libraries and place them in their appropriate locations on the CI server. Next, I had to re-engineer the xvfb-run script. Here is a copy of my script (NOTE: This is the solution for my server and this may not work for you) I created a script that contains: /usr/openwin/bin/Xvfb :1 screen 0 1280x1024x8 pixdepths 8 24 fbdir /tmp/.X11-vbf & From http://www.ensor.cc/2011/08/maven-javascript-unit-test-running-in.html

Approval Tests or simply Approvals is a framework created by Llewellyn Falco and Dan Gilkerson, providing support for .NET, Java, PHP and Ruby. It is not yet another unit testing framework like NUnit or MbUnit etc.; instead these frameworks are used to run approval tests. Broadly speaking, software is nothing more than virtual box there we put in some inputs and expect some outputs. The outputs could be produce in a zillion ways. Those ways differ by implementation. Unit tests are too much focused on implementation. That's why unit tests might fail even if you have working code. Approvals, by contrast, are focused on output. How does it work? Let's take a look at a very simple case. Say, I have a class ShoppingCart. I can add some products inside the shopping cart, and confirm my purchase. I expect that the total price is calculated for me. [TestFixture] [UseReporter(typeof(DiffReporter))] public class ShoppingCartTests { [Test] public void should_calculate_the_total_price_for_shopping_cart() { // do var shoppingCart = new ShoppingCart(); shoppingCart.Add(new Product { Id = "iPad", Price = 500 }); shoppingCart.Add(new Product { Id = "Mouse", Price = 20 }); shoppingCart.Confirm(); // verify Approvals.Approve(shoppingCart); } } What happens if I run this test? If I'm running it the first time, it fails. No matter whether it works or doesn't. The framework simply doesn't know that yet. To understand how correct that code is, use your human power of recognition. In that case, you'll see that it will open the TortoiseDiff application and show the actual and expected outputs. Here, I can tell: "Ok, I have 2 products in my cart..one iPod and one Mouse, iPods costs 500 smth and mouse is 20 smth.. and the total price is 520 - looks good! I approve that result!". Technically, the approving is just copying actual output to expected. As soon as the test passes, the actual file output is deleted and the approved file resides near the test code file, so you just check it in source control. But let's say the shopping cart is modified and something goes wrong. There would be a failure. In the case of unit tests, that would be multiple failures of different cases and it might be not so easy to understand exactly what's wrong. For an approval test, on the other hand, it would be just one failure. And the cooles thing is that can see exactly where the deviation is. Where does it work? It is not only the simple objects that you can approve. You can even approve different sources: objects, enumerables, files, HTML, XML etc. On a more high level: WpfForm, WinForm, ASP.NET Page. For instance, code for ASP.NET: [Test] public void should_have_approved_layout() { ApprovalTests.Asp.Approvals.ApproveUrl("http://localhost:62642/customer/"); } Or for a WPF form: [Test] public void should_have_approved_layout() { ApprovalTests.Wpf.Approvals.Approve(new Form()); } With WPF and Win forms, it's able to serialize them into images, so the actual and expected results are actually images, so it is easy to track the differences (TortoiseDiff can do that). When should you use it? It works best when you deal with 2 things: UI and legacy code. Testing UI is always difficult. But what you typically need to do is: make sure that UI is not changed, and if it has changed, know where exactly the change happened. Approval testing solves that nicely. It takes only one line of code to test ASP.NET page, for instance. Legacy is another story: you have no tests there at all, but you have to change code to implement a new feature, or refactor. The interesting thing about legacy codeis - It works! It works for years, no matter how it is written (remember, virtual box). And this is a very great advantage of that code. With approvals, with only one test you can get all possible outputs (HTML, XLM, JSON, SQL or whatever output it could be) and approve, because you know - it works! After you have complete such a test and approved the result, you are really much safer with a refactoring, since now you "locked down" all existing behavior. Approvals are not something you need to run all the time, like units or integration tests. Approval testing is more like handy tool. You create approval tests, you do your job and at the end of the day it might happen - the tool is no longer needed, so you can just throw the tool away. Want to hear more? Just go and listen to this Herding Code podcast episode, or visit the project web site or join me at 17 December on XP Days Ukraine conference in Kiev, where I'm going to have a speech dedicated to Approvals. Source: http://www.beletsky.net/2011/12/approval-tests-alternative-view-on-test.html

Everyone knows from reading The Mythical Man Month that as you add more people to a software development project you will see diminishing marginal returns. When you add a person to a team, there’s a short-term hit as the rest of the team slows down to bring the new team member up to speed and adjusts to working with another person, making sure that they fit in and can contribute. There’s also a long-term cost. More people means more people who need to talk to each other (n x n-1 / 2), which means more opportunities for misunderstandings and mistakes and misdirections and missed handoffs, more chances for disagreements and conflicts, more bottleneck points. As you continue to add people, the team needs to spend more time getting each new person up to speed and more time keeping everyone on the team in synch. Adding more people means that the team speeds up less and less, while people costs and communications costs and overhead costs keep going up. At some point negative returns set in – if you add more people, the team’s performance will decline and you will get less work done, not more. Diminishing Returns from any One Practice But adding too many people to a project isn’t the only case of diminishing returns in software development. If you work on a big enough project, or if you work in maintenance for long enough, you will run into problems of diminishing returns everywhere that you look. Pushing too hard in one direction, depending too much on any tool or practice, will eventually yield diminishing returns. This applies to: - Manual functional and acceptance testing - Test automation - Any single testing technique - Code reviews - Static analysis bug finding tools - Penetration tests and other security reviews Aiming for 100% code coverage on unit tests is a good example. Building a good automated regression safety net is important – as you wire in tests for key areas of the system, programmers get more confidence and can make more changes faster. How many tests are enough? In Continuous Delivery, Jez Humble and David Farley set 80% coverage as a target for each of automated unit testing, functional testing and acceptance testing. You could get by with lower coverage in many areas, higher coverage in core areas. You need enough tests to catch common and important mistakes. But beyond this point, more tests get more difficult to write, and find fewer problems. Unit testing can only find so many problems in the first place. In Code Complete, Steve McConnell explains that unit testing can only find between 15% and 50% (on average 30%) of the defects in your code. Rather than writing more unit tests, people’s time would be better spent on other approaches like exploratory system testing and code reviews or stress testing or fuzzing to find different kinds of errors. Too much of anything is bad, but too much whiskey is enough. Mark Twain, as quoted in Code Complete Refactoring is important for maintaining and improving the structure and readability of code over time. It is intended to be a supporting practice – to help make changes and fixes simpler and clearer and safer. When refactoring becomes an end in itself or turns into Obsessive Refactoring Disorder, it not only adds unnecessary costs as programmers waste time over trivial details and style issues, it can also add unnecessary risks and create conflict in a team. Make sure that refactoring is done in a disciplined way, and focus refactoring on those areas that need it the most: on code that is frequently changed, routines that are too big, too hard to read, too complex and error-prone. Putting most of your attention refactoring (or if necessary rewriting) this code will get you the highest returns. Less and Less over Time Diminishing returns also set in over time. The longer that you spend working the same way and with the same tools, the less benefits you will see. Even core practices that you’ve grown to depend on don’t pay back over time, and at some point may cost more than they are worth. It’s time again for New Year’s resolutions – time to sign up at a gym and start lifting weights. If you stick with the same routine for a couple of months, you will start to see good results. But after a while your body will get used to the work – if you keep doing the same things the same way your performance will plateau and you will stop seeing gains. You will get bored and stop going to the gym, which will leave more room for people like me. If you do keep going, trying to push harder for returns, you will overtrain and injure yourself. The same thing happens to software teams following the same practices, using the same tools. Some of this is due to inertia. Teams, organizations reach an equilibrium point and they want to stay there. Because it is comfortable, and it works – or at least they understand it. And because the better the team is working, the harder it is to get better – all the low-hanging fruit has been picked. People keep doing what worked for them in the past. They stop looking beyond their established routines, stop looking for new ideas. Competence and control lead to complacency and acceptance. Instead of trying to be as good as possible, they settle for being good enough. This is the point of inspect-and-adapt in Scrum and other time boxed methods – asking the team to regularly re-evaluate what they are doing and how they are doing it, what’s going well and what isn’t, what they should do more of or less of, challenging the status quo and finding new ways to move forward. But even the act of assessing and improving is subject to diminishing returns. If you are building software in 2-week time boxes, and you’ve been doing this for 3, 4 or 5 years, then how much meaningful feedback should you really expect from so many superficial reviews? After a while the team finds themselves going over the same issues and problems and coming up with the same results. Reviews become an unnecessary and empty ritual, another waste of time. The same thing happens with tools. When you first start using a static analysis bug checking tool for example, there’s a good chance that you will find some interesting problems that you didn’t know were in the code – maybe even more problems than you can deal with. But once you triage this and fix up the code and use the tool for a while, the tool will find fewer and fewer problems until it gets to the point where you are paying for insurance – it isn’t finding problems any more, but it might someday. In "Has secure software development reached its limits?” William Jackson argues that SDLCs – all of them – eventually reach a point of diminishing returns from a quality and security standpoint, and that Microsoft and Oracle and other big shops are already seeing diminishing returns from their SDLCs. Their software won’t get any better – all they can do is to keep spending time and money to stay where they are. The same thing happens with Agile methods like Scrum or XP – at some point you’ve squeezed everything that you can from this way or working, and the team’s performance will plateau. What can you do about diminishing returns? First, understand and expect returns to diminish over time. Watch for the signs, and factor this into your expectations – that even if you maintain discipline and keep spending on tools, you will get less and less return for your time and money. Watch for the team’s velocity to plateau or decline. Expect this to happen and be prepared to make changes, even force fundamental changes on the team. If the tools that you are using aren’t giving returns any more, then find new ones, or stop using them and see what happens. Keep reviewing how the team is working, but do these reviews differently: review less often, make the reviews more focused on specific problems, involve different people from inside and outside of the team. Use problems or mistakes as an opportunity to shake things up and challenge the status quo. Dig deep using Root Cause Analysis and challenge the team’s way of thinking and working, look for something better. Don’t settle for simple answers or incremental improvements. Remember the 80/20 rule. Most of your problems will happen in the same small number of areas, from a small number of common causes. And most of your gains will come from a few initiatives. Change the team’s driving focus and key metrics, set new bars. Use Lean methods and Lean Thinking to identify and eliminate bottlenecks, delays and inefficiencies. Look at the controls and tests and checks that you have added over time, question whether you still need them, or find steps and checks that can be combined or automated or simplified. Focus on reducing cycle time and eliminating waste until you have squeezed out what you can. Then change your focus to quality and eliminating bugs, or to simplifying the release and deployment pipeline, or some other new focus that will push the team to improve in a meaningful way. And keep doing this and pushing until you see the team slowing down and results declining. Then start again, and push the team to improve again along another dimension. Keep watching, keep changing, keep moving ahead. Source: http://swreflections.blogspot.com/2011/11/diminishing-returns-in-software.html

For most large web applications, uptime is of foremost importants. Any outage can be seen by customers as a frustration, or opportunity to move to a competitor. What's more for a site that also includes e-commerce, it can mean real lost sales. Zero Downtime describes a site without service interruption. To achieve such lofty goals, redundancy becomes a critical requirement at every level of your infrastructure. If you're using cloud hosting, are you redundant to alternate availability zones and regions? Are you using geographically distributed load balancing? Do you have multiple clustered databases on the backend, and multiple webservers load balanced. All of these requirements will increase uptime, but may not bring you close to zero downtime. For that you'll need thorough testing. The solution is to pull the trigger on sections of your infrastructure, and prove that it fails over quickly without noticeable outage. The ultimate test is the outage itself. Sean Hull on Quora: What is zero downtime and why is it important? Source: http://www.iheavy.com/2011/06/23/zero-downtime-what-is-it-and-why-is-it-important/

A brief overview of dependency injection in Java EE, the difference between @Resource/@EJB and @Inject, and how does that all relate to Spring – mostly in the form of links. Context Dependency Injection (CDI, JSR 299) is a part of Java EE 6 Web Profile and itself builds on Dependency Injection for Java (JSR 330), which introduces @Inject, @Named etc. While JSR 330 is for DI only and is implemented e.g. by Guice and Spring, CDI adds various EE stuff such as @RequestScoped, interceptors/decorators, producers, eventing and a base for integration with JSF, EJBs etc. Java EE components such as EJBs have been redefined to build on top of CDI (=> @Stateless is now a CDI managed bean with additional services). A key part of CDI aside of its DI capabilities is its awarness of bean contexts and the management of bean lifecycle and dependencies within those contexts (such as @RequestScoped or @ConversationScoped). CDI is extensible – you can define new context scopes, drop-in interceptors and decorators, make other beans (e.g. from Spring) available for CDI,… . Resources to check: Contexts and Dependency Injection in Java EE 6 by Adam Bien – a very good explanation of the basics of CDI and how it differs from DI in Java EE 5 (hint: context awarness) Slideshow with a good overview of CDI and all it offers About CDI extensibility and SPIs (e.g. Seam 3 is basically a set of portable CDI extensions) Guice and Spring do not implement CDI (3/2011) – and Spring perhaps isn’t motivated to do so (it supports JSR 330, CDI would be too much work) DZone CDI Refcard may be handy CDI 1.0 vs. Spring 3.1 feature comparsion: bean definition & dependency injection: “in the area that I compared in this article [= DI], there is only little critical difference in the two technologies” (though Spring more fine-tunable) Java EE 6 (CDI / EJB 3.1) XOR Spring Core Reloaded: New projects should preferably start with pure Java EE including CDI and add Spring utilities such as JDBC/JMS when needed Oracle: CDI in the Java EE 6 Ecosystem – 62 pages slideshow, the stuff is explained more than in the previously mentioned slideshow Note: CDI 1.1 (JSR 346, Java EE 7) should have a standard way of bootstrapping it in non-EE environment (i.e. SE) From http://theholyjava.wordpress.com/2011/11/09/what-is-cdi-how-does-it-relate-to-ejb-and-spring/

Dependency injection is a very simple concept: if you have an object that interacts with other objects the responsibility of finding a reference to those objects at run time is moved outside of the object itself. What does it mean for an object to "interact" with other objects? Generally it means invoking methods or reading properties from those objects. So if we have a class A that invokes method Calculate on class B, we can say that A interacts with B. In the following example we show class A interacting with class B. We can equally say that A depends on class B to fulfill a responsibility. In this case, it not only invokes its method Calculate but it also creates a new instance of that class. class A { private B _b; public A { _b = new B(); } public int SomeMethod() { return (_b.Calculate() * 2); } } In the following example, on the other side, the responsibility of getting a reference to an implementation of a class of type B is moved outside of A: class A { private _b = B; public A(B b) { _b = b; } public int SomeMethod() { return _(b.Calculate * 2); } } In this case we say that a dependency (B) has been injected into A, via the constructor. Of course, you can also inject dependencies via a property (or even a regular method), like in the following example: class A { private _b = B; public B B { get { return _b; } set { _b = value; } } public int SomeMethod() { if (_b != null) return _b.RetrieveValue() * 2;} else // HANDLE THIS ERROR CASE return -1; } } So this is all there is about dependency injection. Everything else just builds on this core concept. Like for example Inversion Of Control (IoC) tools which helps you wiring together your objects at run time, injecting all dependencies as needed. So what exactly is Inversion of Control and how does it relate to Dependency Injection (DI)? I like to associate IoC to the Hollywood Principle: "Don't call us, we'll call you". IoC is a design principle where reusable generic code controls the execution of problem-specific code: it is a characteristic of many frameworks, where the application is built extending or customizing a common skeleton; you put your own classes at specific points and the framework will call you when needed. You can use an IoC container as a framework to perform Dependency Injection on your behalf: you tell the container which are the concrete implementation classes for your dependencies and the container will make sure that your constructors or setters will be called with the right objects. Therefore, IoC containers are just a convenience to simplify how dependency injection is handled. But even if you don't use one you could still manually perform dependency injection. (If you want to have a look at how an IoC container works you can jump to my mini tutorial on Ninject). Why is the concept of dependency injection important? Because by applying it, you simplify your design (separating the responsibility of using an object from the responsibility of using the object) and your code becomes much easier to test, since you can mock out the dependencies substituting them with fake (stub) objects. But that is the subject for another post.

In my opinion, one of the main benefits of dependency injection is that you can inject mock and/or stub objects into your code in order to improve testability, increase test coverage and write better and more meaningful tests. There are those times, however, when you come across some legacy code that doesn’t use dependency injection and held together by composition rather than aggregation. When this happens, you have three options: Ignore the problem and not write any tests. Refactor like mad, changing everything to use dependency injection. Use PowerMock to mock constructors Obviously, option 1 isn’t a serious option, and although I’d recommend refactoring to move everything over to dependency injection, that takes time and you have to be pragmatic. That’s where PowerMock comes in... this blog demonstrates how to use PowerMock to mock a constructor, which means that when your code calls new it doesn’t create a real object, it creates a mock object. To demonstrate this idea, the first thing we need is some classes to test, which are shown below. public class AnyOldClass { public String someMethod() { return "someMethod"; } } public class UsesNewToInstantiateClass { public String createThing() { AnyOldClass myclass = new AnyOldClass(); String returnValue = myclass.someMethod(); return returnValue; } } The first class, AnyOldClass, is the class that the code instantiates by calling new. In this example, as the name suggests, it can be anything. The second class, the aptly named UsesNewToInstantiateClass, has one method, createThing(), which when called does a: AnyOldClass myclass = new AnyOldClass(); This is all pretty straight forward, so we’ll move quickly on to the PowerMock assisted JUnit test: import static org.easymock.EasyMock.expect; import static org.junit.Assert.assertEquals; import static org.powermock.api.easymock.PowerMock.expectNew; import static org.powermock.api.easymock.PowerMock.replay; import static org.powermock.api.easymock.PowerMock.verify; import org.junit.Test; import org.junit.runner.RunWith; import org.powermock.api.easymock.annotation.Mock; import org.powermock.core.classloader.annotations.PrepareForTest; import org.powermock.modules.junit4.PowerMockRunner; @RunWith(PowerMockRunner.class) @PrepareForTest(UsesNewToInstantiateClass.class) public class MockConstructorTest { @Mock private AnyOldClass anyClass; private UsesNewToInstantiateClass instance; @Test public final void testMockConstructor() throws Exception { instance = new UsesNewToInstantiateClass(); expectNew(AnyOldClass.class).andReturn(anyClass); final String expected = "MY_OTHER_RESULT"; expect(anyClass.someMethod()).andReturn(expected); replay(AnyOldClass.class, anyClass); String result = instance.createThing(); verify(AnyOldClass.class, anyClass); assertEquals(expected, result); } } Firstly, this class has the usual PowerMock additions of: @RunWith(PowerMockRunner.class) @PrepareForTest(UsesNewToInstantiateClass.class) at the top of the file plus the creation of the anyOldClass mock object. The important line of code to consider is: expectNew(AnyOldClass.class).andReturn(anyClass); This line of code tells PowerMock to expect a call to new AnyOldClass() and return our anyClass mock object. Also of interest are the calls to replay and verify. In the example above, they both have two arguments. The first, AnyOldClass.class relates to the expectNew(...) call above, whilst the second, anyClass refers to the straight forward mock call expect(anyClass.someMethod()).andReturn(expected);. There are those times when you should really let new do what it does: create a new object of the requested type. There is a body of opinion that says you can over-isolate your code when testing and that mocking everything reduces the meaning and value of a test. To me there’s no right answer to this and it’s a matter of choice. It’s fairly obvious that if your code accesses an external resource such as a database, then you’d either refactor and implement DI or use PowerMock. If your code under test doesn’t access any external resources, then it’s more of a judgement call on how much code isolation is too much? This perhaps needs some thought and may be the subject for another blog on anther day... From http://www.captaindebug.com/2011/10/using-powermock-to-mock-constructors.html

One of the challenges in building a distributed web application is in handling sessions. When you have multiple instances of an application running and session data is written to local files (as is the default behavior for the session handling functions in PHP) a user session can be lost when a session is started on one instance but subsequent requests are directed (via a load balancer) to other instances. To successfully manage sessions across multiple instances, you need a common data store. In this post I’ll show you how the Windows Azure SDK for PHP makes this easy by storing session data in Windows Azure Table storage. In the 4.0 release of the Windows Azure SDK for PHP, session handling via Windows Azure Table and Blob storage was included in the newly added SessionHandler class. Note: The SessionHandler class supports storing session data in Table storage or Blob storage. I will focus on using Table storage in this post largely because I haven’t been able to come up with a scenario in which using Blob storage would be better (or even necessary). If you have ideas about how/why Blob storage would be better, I’d love to hear them. The SessionHandler class makes it possible to write code for handling sessions in the same way you always have, but the session data is stored on a Windows Azure Table instead of local files. To accomplish this, precede your usual session handling code with these lines: require_once 'Microsoft/WindowsAzure/Storage/Table.php'; require_once 'Microsoft/WindowsAzure/SessionHandler.php'; $storageClient = new Microsoft_WindowsAzure_Storage_Table('table.core.windows.net', 'your storage account name', 'your storage account key'); $sessionHandler = new Microsoft_WindowsAzure_SessionHandler($storageClient , 'sessionstable'); $sessionHandler->register(); Now you can call session_start() and other session functions as you normally would. Nicely, it just works. Really, that’s all there is to using the SessionHandler, but I found it interesting to take a look at how it works. The first interesting thing to note is that the register method is simply calling the session_set_save_handler function to essentially map the session handling functionality to custom functions. Here’s what the method looks like from the source code: public function register() { return session_set_save_handler(array($this, 'open'), array($this, 'close'), array($this, 'read'), array($this, 'write'), array($this, 'destroy'), array($this, 'gc') ); } The reading, writing, and deleting of session data is only slightly more complicated. When writing session data, the key-value pairs that make up the data are first serialized and then base64 encoded. The serialization of the data allows for lots of flexibility in the data you want to store (i.e. you don’t have to worry about matching some schema in the data store). When storing data in a table, each entry must have a partition key and row key that uniquely identify it. The partition key is a string (“sessions” by default, but this is changeable in the class constructor) and the the row key is the session ID. (For more information about the structure of Tables, see this post.) Finally, the data is either updated (it it already exists in the Table) or a new entry is inserted. Here’s a portion of the write function: $serializedData = base64_encode(serialize($serializedData)); $sessionRecord = new Microsoft_WindowsAzure_Storage_DynamicTableEntity($this->_sessionContainerPartition, $id); $sessionRecord->sessionExpires = time(); $sessionRecord->serializedData = $serializedData; try { $this->_storage->updateEntity($this->_sessionContainer, $sessionRecord); } catch (Microsoft_WindowsAzure_Exception $unknownRecord) { $this->_storage->insertEntity($this->_sessionContainer, $sessionRecord); } Not surprisingly, when session data is read from the table, it is retrieved by session ID, base64 decoded, and unserialized. Again, here’s a snippet that show’s what is happening: $sessionRecord = $this->_storage->retrieveEntityById( $this->_sessionContainer, $this->_sessionContainerPartition, $id ); return unserialize(base64_decode($sessionRecord->serializedData)); As you can see, the SessionHandler class makes good use of the storage APIs in the SDK. To learn more about the SessionHandler class (and the storage APIs), check out the documentation on Codeplex. You can, of course, get the complete source code here: http://phpazure.codeplex.com/SourceControl/list/changesets. As I investigated the session handling in the Windows Azure SDK for PHP, I noticed that the absence of support for SQL Azure as a session store was conspicuous. I’m curious about how many people would prefer to use SQL Azure over Azure Tables as a session store. If you have an opinion on this, please let me know in the comments.

I’ve been going over a couple of posts by Steve Kilner that question whether Agile methods can be used effectively in software maintenance. It’s a surprising question really. There are a lot of maintenance teams who have had success following Agile methods like Scrum and Extreme Programming (XP) for some time now. We’ve been doing it for almost 5 years, enhancing and maintaining and supporting enterprise systems, and I know that it works. Agile development naturally leads into maintenance – the goal of incremental Agile development is to get working software out to customers as soon as possible, and get customers using it. At some point, when customers are relying on the software to get real business done and need support and help to keep the system running, teams cross from development over to maintenance. But there’s no reason for Agile development teams to fundamentally change the way that they work when this happens. It is harder to introduce Agile practices into a legacy maintenance team – there are a lot of technical requirements and some cultural changes that need to be made. But most maintenance teams have little to lose and lots to gain from borrowing from what Agile development teams are doing. Agile methods are designed to help small teams deal with a lot of change and uncertainty, and to deliver software quickly – all things that are at least as important in maintenance as they are in development. Technical practices in Extreme Programming especially help ensure that the code is always working – which is even more important in maintenance than it is in development, because the code has to work the first time in production. Agile methods have to be adapted to maintenance, but most teams have found it necessary to adapt these methods to fit their situations anyways. Let’s look at what works and what has to be changed to make Agile methods like Scrum and XP work in maintenance. What works well and what doesn’t Planning Game Managing maintenance isn’t the same as managing a development project – even an Agile development project. Although Agile development teams expect to deal with ambiguity and constant change, maintenance teams need to be even more flexible and responsive, to manage conflicts and unpredictable resourcing problems. Work has to be continuously reviewed and prioritized as it comes in – the customer can’t wait for 2 weeks for you to look at a production bug. The team needs a fast path for urgent changes and especially for hot fixes. You have to be prepared for support demands and interruptions. Structure the team so that some people can take care of second-level support, firefighting and emergency bug fixing and the rest of the team can keep moving forward and get something done. Build slack into schedules to allow for last-minute changes and support escalation. You will also have to be more careful in planning out maintenance work, to take into account technical and operational dependencies and constraints and risks. You’re working in the real world now, not the virtual reality of a project. Standups Standups play an important role in Agile projects to help teams come up to speed and bond. But most maintenance teams work fine without standups – since a lot of maintenance work can be done by one person working on their own, team members don’t need to listen to each other each morning talking about what they did yesterday and what they’re going to do – unless the team is working together on major changes. If someone has a question or runs into a problem, they can ask for help without waiting until the next day. Small releases Most changes and fixes that maintenance teams need to make are small, and there is almost always pressure from the business to get the code out as soon as it is ready, so an Agile approach with small and frequent releases makes a lot of sense. If the time boxes are short enough, the customer is less likely to interrupt and re-prioritize work in progress – most businesses can wait a few days or a couple of weeks to get something changed. Time boxing gives teams a way to control and structure their work, an opportunity to batch up related work to reduce development and testing costs, and natural opportunities to add in security controls and reviews and other gates. It also makes maintenance work more like a project, giving the team a chance to set goals and to see something get done. But time boxing comes with overhead – the planning and setup at the start, then deployment and reviews at the end – all of which adds up over time. Maintenance teams need to be ruthless with ceremonies and meetings, pare them down, keep only what’s necessary and what works. It’s even more important in maintenance than in development to remember that the goal is to deliver working code at the end of each time box. If some code is not working, or you’re not sure if it is working, then extend the deadline, back some of the changes out, or pull the plug on this release and start over. Don’t risk a production failure in order to hit an arbitrary deadline. If the team is having problems fitting work into time boxes, then stop and figure out what you’re doing wrong – the team is trying to do too much too fast, or the code is too unstable, or people don’t understand the code enough – and fix it and move on. Reviews and Retrospectives Retrospectives are important in maintenance to keep the team moving forward, to find better ways of working, and to solve problems. But like many practices, regular reviews reach a point of diminishing returns over time – people end up going through the motions. Once the team is setup, reviews don’t need to be done in each iteration unless the team runs into problems. Schedule reviews when you or the team need them. Collect data on how the team is working, on cycle time and bug report/fix ratios, correlate problems in production with changes, and get the team together to review if the numbers move off track. If the team runs into a serious problem like a major production failure, then get to the bottom of it through Root Cause Analysis. Sustainable pace / 40-hour week It’s not always possible to work a 40-hour week in maintenance. There are times when the team will be pushed to make urgent changes, spend late nights firefighting, releasing after hours and testing on weekends. But if this happens too often or goes on too long the team will burn out. It’s critical to establish a sustainable pace over the long term, to treat people fairly and give them a chance to do a good job. Pairing Pairing is hard to do in small teams where people are working on many different things. Pairing does make sense in some cases – people naturally pair-up when trying to debug a nasty problem or walking through a complicated change – but it’s not necessary to force it on people, and there are good reasons not to. Some teams (like mine) rely more on code reviews instead of pairing, or try to get developers to pair when first looking at a problem or change, and at the end again to review the code and tests. The important thing is to ensure that changes get looked at by at least one other person if possible, however this gets done. Collective Code Ownership Because maintenance teams are usually small and have to deal with a lot of different kinds of work, sooner or later different people will end up working on different parts of the code. It’s necessary, and it’s a good thing because people get a chance to learn more about the system and work with different technologies and on different problems. But there’s still a place for specialists in maintenance. You want the people who know the code the best to make emergency fixes or high-risk changes – or at least have them review the changes – because it has to work the first time. And sometimes you have no choice – sometimes there is only one person who understands a framework or language or technical problem well enough to get something done. Coding Guidelines – follow the rules Getting the team to follow coding guidelines is important in maintenance to help ensure the consistency and integrity of the code base over time – and to help ensure software security. Of course teams may have to compromise on coding standards and style conventions, depending on what they have inherited in the code base; and teams that maintain multiple systems will have to follow different guidelines for each system. Metaphor In XP, teams are supposed to share a Metaphor: a simple high-level expression of the system architecture (the system is a production line, or a bill of materials) and common names and patterns that can be used to describe the system. It’s a fuzzy concept at best, a weak substitute for more detailed architecture or design, and it’s not of much practical value in maintenance. Maintenance teams have to work with the architecture and patterns that are already in place in the system. What is important is making sure that the team has a common understanding of these patterns and the basic architecture so that the integrity isn’t lost – if it hasn’t been lost already. Getting the team together and reviewing the architecture, or reverse-engineering it, making sure that they all agree on it and documenting it in a simple way is important especially when taking over maintenance of a new system and when you are planning major changes. Simple Design Agile development teams start with simple designs and try to keep them simple. Maintenance teams have to work with whatever design and architecture that they inherit, which can be overwhelmingly complex, especially in bigger and older systems. But the driving principle should still be to design changes and new features as simple as the existing system lets you – and to simplify the system’s design further whenever you can. Especially when making small changes, simple, just-enough design is good – it means less documentation and less time and less cost. But maintenance teams need to be more risk adverse than development teams – even small mistakes can break compatibility or cause a run-time failure or open a security hole. This means that maintainers can’t be as iterative and free to take chances, and they need to spend more time upfront doing analysis, understanding the existing design and working through dependencies, as well as reviewing and testing their changes for regressions afterwards. Refactoring Refactoring takes on a lot of importance in maintenance. Every time a developer makes a change or fix they should consider how much refactoring work they should do and can do to make the code and design clearer and simpler, and to pay off technical debt. What and how much to refactor depends on what kind of work they are doing (making a well-thought-out isolated change, or doing shotgun surgery, or pushing out an emergency hot fix) and the time and risks involved, how well they understand the code, how good their tools are (development IDEs for Java and .NET at least have good built-in tools that make many refactorings simple and safe) and what kind of safety net they have in place to catch mistakes – automated tests, code reviews, static analysis. Some maintenance teams don’t refactor because they are too afraid of making mistakes. It’s a vicious circle – over time the code will get harder and harder to understand and change, and they will have more reasons to be more afraid. Others claim that a maintenance team is not working correctly if they don’t spend at least 50% of their time refactoring. The real answer is somewhere in between – enough refactoring to make changes and fixes safe. There are cases where extensive refactoring, restructuring or rewriting code is the right thing to do. Some code is too dangerous to change or too full of bugs to leave the way it is – studies show that in most systems, especially big systems, 80% of the bugs can cluster in 20% of the code. Restructuring or rewriting this code can pay off quickly, reducing problems in production, and significantly reducing the time needed to make changes and test them as you go forward. Continuous Testing Testing is even more important and necessary in maintenance than it is in development. And it’s a major part of maintenance costs. Most maintenance teams rely on developers to test their own changes and fixes by hand to make sure that the change worked and that they didn’t break anything as a side effect. Of course this makes testing expensive and inefficient and it limits how much work the team can do. In order to move fast, to make incremental changes and refactoring safe, the team needs a better safety net, by automating unit and functional tests and acceptance tests. It can take a long time to put in test scaffolding and tools and write a good set of automated tests. But even a simple test framework and a small set of core fat tests can pay back quickly in maintenance, because a lot changes (and bugs) tend to be concentrated in the same parts of the code – the same features, framework code and APIs get changed over and over again, and will need to be tested over and over again. You can start small, get these tests running quickly and reliably and get the team to rely on them, fill in the gaps with manual tests and reviews, and then fill out the tests over time. Once you have a basic test framework in place, developers can take advantage of TFD/TDD especially for bug fixes – the fix has to be tested anyways, so why not write the test first and make sure that you fixed what you were supposed to? Continuous Integration To get Continuous Testing to work, you need a Continuous Integration environment. Understanding, automating and streamlining the build and getting the CI server up and running and wiring in tests and static analysis checks and reporting can take a lot of work in an enterprise system, especially if you have to deal with multiple languages and platforms and dependencies between systems. But doing this work is also the foundation for simplifying release and deployment – frequent short releases means that release and deployment has to be made as simple as possible. Onsite Customer / Product Owner Working closely with the customer to make sure that the team is delivering what the customer needs when the customer needs it is as important in maintenance as it is in developing a new system. Getting a talented and committed Customer engaged is hard enough on a high-profile development project – but it’s even harder in maintenance. You may end up with too many customers with conflicting agendas competing for the team’s attention, or nobody who has the time or ability to answer questions and make decisions. Maintenance teams often have to make compromises and help fill in this role on their own. But it doesn’t all fit…. Kilner’s main point of concern isn’t really with Agile methods in maintenance. It’s with incremental design and development in general – that some work doesn’t fit nicely into short time boxes. Short iterations might work ok for bug fixes and small enhancements (they do), but sometimes you need to make bigger changes that have lots of dependencies. He argues that while Agile teams building new systems can stub out incomplete work and keep going in steps, maintenance teams have to get everything working all at once – it’s all or nothing. It’s not easy to see how big changes can be broken down into small steps that can be fit into short time boxes. I agree that this is harder in maintenance because you have to be more careful in understanding and untangling dependencies before you make changes, and you have to be more careful not to break things. The code and design will sometimes fight the kinds of changes that you need to make, because you need to do something that was never anticipated in the original design, or whatever design there was has been lost over time and any kind of change is hard to make. It’s not easy – but teams solve these problems all the time. You can use tools to figure out how much of a dependency mess you have in the code and what kind of changes you need to make to get out of this mess. If you are going to spend “weeks, months, or even years” to make changes to a system, then it makes sense to take time upfront to understand and break down build dependencies and isolate run-time dependencies, and put in test scaffolding and tests to protect the team from making mistakes as they go along. All of this can be done in time boxed steps. Just because you are following time boxes and simple, incremental design doesn’t mean that you start making changes without thinking them through. Read Working With Legacy Code – Michael Feathers walks through how to deal with these problems in detail, in both object oriented and procedural languages. What to do if it takes forever to make a change. How to break dependencies. How to find interception points and pinch points. How to find structure in the design and the code. What tests to write and how to get automated tests to work. Changing data in a production system, especially data shared with other systems, isn’t easy either. You need to plan out API changes and data structure changes as carefully as possible, but you can still make data and database changes in small, structured steps. To make code changes in steps you can use Branching by Abstraction where it makes sense (like making back-end changes) and you can protect customers from changes through Feature Flags and Dark Launching like Facebook and Twitter and Flickr do to continuously roll out changes – although you need to be careful, because if taken too far these practices can make code more fragile and harder to work with. Agile development teams follow incremental design and development to help them discover an optimal solution through trial-and-error. Maintenance teams work this way for a different reason – to manage technical risks by breaking big changes down and making small bets instead of big ones. Working this way means that you have to put in scaffolding (and remember to take it out afterwards) and plan out intermediate steps and review and test everything as you make each change. Sometimes it might feel like you are running in place, that it is taking longer and costing more. But getting there in small steps is much safer, and gives you a lot more control. Teams working on large legacy code bases and old technology platforms will have a harder time taking on these ideas and succeeding with them. But that doesn’t mean that they won’t work. Yes, you can be Agile in maintenance.