How to use XmlSuiteUtils class of org.testng.xml.internal package

That kind of dynamic behaviour isn't supported out-of-the-box, but you can achieve a lot with module rewriting. Take a look at Guiceberry, which already implements mock-substitution for JUnit tests. (And consider submitting a TestNG patch to them, they'd love that!)

In Intellij, go to "edit configurations" (Run -> Edit Configurations), in the configuration tab that comes up in the line "shorten command line" select option "classpath file"/"@argFiles (Java 9+)", last option in the drop down menu. Fixed it for me.

IMHO, yes they are still relevant.

With EJB3, you can test you EJB either like regular POJO, or as managed bean using an embedded EJB container.

Same for JPA, you can either embed a JPA implementation for your test and use an in-memory database, or you can mock the data layer completely.

For my last EJB project I had written a bit of glue code to test the EJB because embedded EJB container were not mature enough, but now I would go for an embedded EJB container + JUnit.

A few resources

• Blog post: How to unit test EJB 3 in 0.8 seconds
• SO question: Good unit testing resource for EJB

But there are many others easily findable

I just had this problem and I'm pretty confident that it's a bug in the TestNG Plugin for eclipse, since I was able to run my tests from the command line just fine.

The fix that seems to be working for me and others is just to clean your project (Project → Clean). One person had to re-install the whole plugin.

You should read entire file in data provider and return iterator of test cases. Here is some pseudocode for data provider. Notice that I used List<String []> to store test cases instead of Object[][]. This allows you do define test cases dynamically.

    @DataProvider(name = "test")
    public Iterator<Object []> provider( ) throws InterruptedException
    {
        List<Object []> testCases = new ArrayList<>();
        String[] data= null;

        //this loop is pseudo code
        br = new BufferedReader(new FileReader(csvFile));
        while ((line = br.readLine()) != null) {
            // use comma as separator
            data= line.split(cvsSplitBy);
            testCases.add(data);
        }

        return testCases.iterator();
    }

I encountered a problem with similar symptoms today. I'm not sure if it is the same problem but it matches everything in your question.

In my case, the log files said:

!ENTRY org.eclipse.core.jobs 4 2 2015-12-03 15:49:19.369
!MESSAGE An internal error occurred during: "Launching NewTest".
!STACK 0
java.lang.NullPointerException
at org.testng.eclipse.maven.MavenTestNGLaunchConfigurationProvider.getVMArgsFromPom(MavenTestNGLaunchConfigurationProvider.java:74)

So my problem was that the TestNG-plugin in Eclipse failed to find the pom-file, as I don't use Maven. Anyway, after uninstalling the optional M2E (Maven...) part of the TestNG Eclipse plugin I could run my TestNG test cases successfully.

If you have the same issue, you uninstall the M2E part in Eclipse by selecting

• Help
• Install new software...
• What is already installed?
• Select the TestNG M2E software and click uninstall

Hope this helps!

Selenium Grid

Selenium Grid was built to allow multiple selenium sessions to be ran and handled automatically. It is configurable for multiple devices (when used in conjunction with appium), allows for parallel or standard testing and allows you to create nodes for each device based on a capability system.

Hub

Selenium Grid hub is a central server which manages the ability to run multiple tests.

1. Download the Selenium Standalone Server

2. Run the following to start the hub:

   $java -jar path/to/selenium-server-standalone-X.X.X.jar -role hub -port 4444 
   

Node

1. Install node.js

2. Install appium (unless you already have)

   $npm install –g appium
   

3. You will need a JSON file for each node, which outlines the capabilities of the device (or devices) that this node is to be used for:

{
"capabilities":
[
{
"browserName": "Chrome",
"deviceName": "deviceName",
"device": "device",
"udid":"XXXXXXXX",
"version":"X.X.X",
"maxInstances": 5,
"platform":"ANDROID",
"platformName": "Android"
}
],
"configuration":
{
"cleanUpCycle":2000,
"timeout":10000,
"proxy": "org.openqa.grid.selenium.proxy.DefaultRemoteProxy", 
"url":"http://0.0.0.0:4723/wd/hub",
"maxSession": 5,
"port": 4723,
"host": "0.0.0.0",
"register": true,
"registerCycle": 5000,
"hubPort": 4444,
"hubHost": "localhost"
}
}

Keep in mind each node needs to use a different port for the url and port configurations. 4723, for example.

4. Create a node

   $appium --nodeconfig /Path/To/Config.json -p 4723 -bp 5723 --chromedriver-port 6723
   

Each node needs three unique ports (Port, Bootstrap Port, ChromeDriver Port) and the port specified with -p I found needs to be the same as the port given in the config.

Testing

I would recommend using a testing framework such as TestNG or JUnit, which will allow you to run test suites and pass capabilities in @BeforeClass methods etc.

But essentially all you need to do is create a AndroidDriver using capabilities identical to those specified in your JSON config file for a certain node and then that instance of AndroidDriver will be associated with that specific node (Selenium Grid handles this) and furthermore with the device which the node is associated with. It's very simple really, once you know what you are doing!

Initialise AndroidDriver like this:

driver.set(new AndroidDriver(new URL(SELENIUM_HUB_URL),
                capabilities)); 

A common mistake people make is passing the AndroidDriver the node URL instead of the Hub URL (http://0.0.0.0:4723/wd/hub instead of http://0.0.0.0:4444/wd/hub) this is really common because it will, at least initially work, but then when trying to use any Selenium Grid specific functionality (such as multiple devices) everything will break.

Good luck and I hope this helps,

Liam

I can think of several options:

• Finding out how much memory your method requires via a microbenchmark (i.e. jmh).
• Building allocation strategies based on heuristic estimation. There are several open source solutions implementing class size estimation i.e. ClassSize. A much easier way could be utilizing a cache which frees rarely used objects (i.e. Guava's Cache). As mentioned by @EnnoShioji, Guava's cache has memory-based eviction policies.

You can also write your own benchmark test which counts memory. The idea is to

1. Have a single thread running.
2. Create a new array to store your objects to allocate. So these objects won't be collected during GC run.
3. System.gc(), memoryBefore = runtime.totalMemory() - runtime.freeMemory()
4. Allocate your objects. Put them into the array.
5. System.gc(), memoryAfter = runtime.totalMemory() - runtime.freeMemory()

This is a technique I used in my lightweight micro-benchmark tool which is capable of measuring memory allocation with byte-precision.

Don't provide priority and depends on together, you can group the tests. You can do it like :

@Test(priority = 1, groups = { "qty" })
    public void increaseQty() {
    System.out.println("in increase qty");
    }

    @Test(dependsOnMethods = { "increaseQty" }, groups = { "qty" })
    public void decreaseQty() {
    System.out.println("in decrease qty");
    }

    @Test(dependsOnGroups = { "qty" })
    public void removeFromCart() throws Exception {
    System.out.println("remove qty");
    }

    @Test(dependsOnMethods = { "removeFromCart" })
    public void emptyCart() throws InterruptedException {
    System.out.println("empty Cart");
    }

Edit: Another work around is to change your testng xml to have desciption like:

<methods> <include name="increaseQty"/> <include name="decreaseQty"/> <include name="removeFromCart"/> <include name="emptyCart"/> </methods>

so testng will execute it in this manner only.