Running Tests in Parallel with Junit [duplicate] - java

This question already has answers here:
Running junit tests in parallel in a Maven build?
(10 answers)
Closed 7 years ago.
I would like to run every method annotated with #Test, across multiple classes, at the same time. For some cases, I would like to limit this, and say that only 100 total can run at any one time. I would like methods with #BeforeClass annotations to be run once before any Test in a class runs, and I would like #AfterClass annotations to be run once after all Tests in a class run. I would like System.out, System.err, and Exceptions to be appropriately buffered/captured rather than written out, so that they don't interleave, and I can read the final output and understand what happened.
Does this exist? I have a large number of independent test cases, and my application is (I believe) threadsafe. None of these tests have dependencies out of the JVM, and I want to finish them as quickly as possible, given my hardware.
If this doesn't exist, is there a concrete reason why not? How much time is lost by junit users worldwide because this isn't easy? Can I build it into Junit? In my mind, this should be as simple as a single flag, and it "just works".

You can accomplish this with JUnit's ParallelComputer (note it's still considered experimental). It's a pretty simple implementation which is backed by the java.util.concurrent.ExecutorService API. If you're curious how it works, check out the source.
Basically you call JUnitCore.runClasses(Computer, Classes ...) and pass in a ParallelComputer object for the first argument.
Example usage:
import org.junit.Test;
import org.junit.experimental.ParallelComputer;
import org.junit.runner.JUnitCore;
public class ParallelComputerExample {
#Test
public void runAllTests() {
Class<?>[] classes = { ParallelTest1.class, ParallelTest2.class };
// ParallelComputer(true,true) will run all classes and methods
// in parallel. (First arg for classes, second arg for methods)
JUnitCore.runClasses(new ParallelComputer(true, true), classes);
}
public static class ParallelTest1 {
#Test
public void test1a() {
lookBusy(3000);
}
#Test
public void test1b() {
lookBusy(3000);
}
}
public static class ParallelTest2 {
#Test
public void test2a() {
lookBusy(3000);
}
#Test
public void test2b() {
lookBusy(3000);
}
}
public static void lookBusy(long ms) {
try {
Thread.sleep(ms);
} catch (InterruptedException e) {
System.out.println("interrupted");
}
}
}
The above code will run in 3 seconds because all methods and classes are ran in parallel.
This will run in 6s (because all classes are in parallel).
JUnitCore.runClasses(new ParallelComputer(true, false), classes);
This will also run in 6s (because all methods are in parallel).
JUnitCore.runClasses(new ParallelComputer(false, true), classes);

Yes, You can.
If you are using maven. You can take help of
maven-surefire-plugin
In Spring,
You can check this Link
<build>
<plugins>
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-surefire-plugin</artifactId>
<version>2.7.1</version>
<configuration>
<parallel>classes</parallel>
<threadCount>5</threadCount>
</configuration>
</plugin>
</plugins>
</build>
Solution 2: Junit4 provides parallel feature using ParallelComputer

JUnit Toolbox provides JUnit runners for parallel execution of tests.
In order to not intermix output to System.err and System.out you have to start tests in separate JVMs, because System.err and System.out are global.

Related

Java code is not printing from start to end [duplicate]

I want to execute test methods which are annotated by #Test in specific order.
For example:
public class MyTest {
#Test public void test1(){}
#Test public void test2(){}
}
I want to ensure to run test1() before test2() each time I run MyTest, but I couldn't find annotation like #Test(order=xx).
I think it's quite important feature for JUnit, if author of JUnit doesn't want the order feature, why?
I think it's quite important feature for JUnit, if author of JUnit doesn't want the order feature, why?
I'm not sure there is a clean way to do this with JUnit, to my knowledge JUnit assumes that all tests can be performed in an arbitrary order. From the FAQ:
How do I use a test fixture?
(...) The ordering of test-method invocations is not guaranteed, so testOneItemCollection() might be executed before testEmptyCollection(). (...)
Why is it so? Well, I believe that making tests order dependent is a practice that the authors don't want to promote. Tests should be independent, they shouldn't be coupled and violating this will make things harder to maintain, will break the ability to run tests individually (obviously), etc.
That being said, if you really want to go in this direction, consider using TestNG since it supports running tests methods in any arbitrary order natively (and things like specifying that methods depends on groups of methods). Cedric Beust explains how to do this in order of execution of tests in testng.
If you get rid of your existing instance of Junit, and download JUnit 4.11 or greater in the build path, the following code will execute the test methods in the order of their names, sorted in ascending order:
import org.junit.FixMethodOrder;
import org.junit.Test;
import org.junit.runners.MethodSorters;
#FixMethodOrder(MethodSorters.NAME_ASCENDING)
public class SampleTest {
#Test
public void testAcreate() {
System.out.println("first");
}
#Test
public void testBupdate() {
System.out.println("second");
}
#Test
public void testCdelete() {
System.out.println("third");
}
}
Migration to TestNG seems the best way, but I see no clear solution here for jUnit. Here is most readable solution / formatting I found for jUnit:
#FixMethodOrder( MethodSorters.NAME_ASCENDING ) // force name ordering
public class SampleTest {
#Test
void stage1_prepareAndTest(){};
#Test
void stage2_checkSomething(){};
#Test
void stage2_checkSomethingElse(){};
#Test
void stage3_thisDependsOnStage2(){};
#Test
void callTimeDoesntMatter(){}
}
This ensures stage2 methods are called after stage1 ones and before stage3 ones.
P.S. I feel this approach is better that jUnit 5.5 #Order annotation because it provides better notation for reader.
If the order is important, you should make the order yourself.
#Test public void test1() { ... }
#Test public void test2() { test1(); ... }
In particular, you should list some or all possible order permutations to test, if necessary.
For example,
void test1();
void test2();
void test3();
#Test
public void testOrder1() { test1(); test3(); }
#Test(expected = Exception.class)
public void testOrder2() { test2(); test3(); test1(); }
#Test(expected = NullPointerException.class)
public void testOrder3() { test3(); test1(); test2(); }
Or, a full test of all permutations:
#Test
public void testAllOrders() {
for (Object[] sample: permute(1, 2, 3)) {
for (Object index: sample) {
switch (((Integer) index).intValue()) {
case 1: test1(); break;
case 2: test2(); break;
case 3: test3(); break;
}
}
}
}
Here, permute() is a simple function which iterates all possible permuations into a Collection of array.
JUnit since 5.5 allows #TestMethodOrder(OrderAnnotation.class) on class and #Order(1) on test-methods.
JUnit old versions allow test methods run ordering using class annotations:
#FixMethodOrder(MethodSorters.NAME_ASCENDING)
#FixMethodOrder(MethodSorters.JVM)
#FixMethodOrder(MethodSorters.DEFAULT)
By default test methods are run in alphabetical order. So, to set specific methods order you can name them like:
a_TestWorkUnit_WithCertainState_ShouldDoSomething
b_TestWorkUnit_WithCertainState_ShouldDoSomething
c_TestWorkUnit_WithCertainState_ShouldDoSomething
Or
_1_TestWorkUnit_WithCertainState_ShouldDoSomething
_2_TestWorkUnit_WithCertainState_ShouldDoSomething
_3_TestWorkUnit_WithCertainState_ShouldDoSomething
You can find examples here.
Its one of the main issue which I faced when I worked on Junit and I came up with following solution which works fine for me:
import java.util.ArrayList;
import java.util.Collections;
import java.util.Comparator;
import java.util.List;
import org.junit.runners.BlockJUnit4ClassRunner;
import org.junit.runners.model.FrameworkMethod;
import org.junit.runners.model.InitializationError;
public class OrderedRunner extends BlockJUnit4ClassRunner {
public OrderedRunner(Class<?> clazz) throws InitializationError {
super(clazz);
}
#Override
protected List<FrameworkMethod> computeTestMethods() {
List<FrameworkMethod> list = super.computeTestMethods();
List<FrameworkMethod> copy = new ArrayList<FrameworkMethod>(list);
Collections.sort(copy, new Comparator<FrameworkMethod>() {
#Override
public int compare(FrameworkMethod f1, FrameworkMethod f2) {
Order o1 = f1.getAnnotation(Order.class);
Order o2 = f2.getAnnotation(Order.class);
if (o1 == null || o2 == null) {
return -1;
}
return o1.order() - o2.order();
}
});
return copy;
}
}
also create a interface like below:
#Retention(RetentionPolicy.RUNTIME)
#Target({ ElementType.METHOD})
public #interface Order {
public int order();
}
Now suppose you have class A where you have written several test cases like below:
(#runWith=OrderRunner.class)
Class A{
#Test
#Order(order = 1)
void method(){
//do something
}
}
So execution will start from method named "method()".
Thanks!
JUnit 5 update (and my opinion)
I think it's quite important feature for JUnit, if author of JUnit doesn't want the order feature, why?
By default, unit testing libraries don't try to execute tests in the order that occurs in the source file.
JUnit 5 as JUnit 4 work in that way. Why? Because if the order matters it means that some tests are coupled between them and that is undesirable for unit tests.
So the #Nested feature introduced by JUnit 5 follows the same default approach.
But for integration tests, the order of the test method may matter since a test method may change the state of the application in a way expected by another test method.
For example when you write an integration test for a e-shop checkout processing, the first test method to be executed is registering a client, the second is adding items in the basket and the last one is doing the checkout. If the test runner doesn't respect that order, the test scenario is flawed and will fail.
So in JUnit 5 (from the 5.4 version) you have all the same the possibility to set the execution order by annotating the test class with #TestMethodOrder(OrderAnnotation.class) and by specifying the order with #Order(numericOrderValue) for the methods which the order matters.
For example :
#TestMethodOrder(OrderAnnotation.class)
class FooTest {
#Order(3)
#Test
void checkoutOrder() {
System.out.println("checkoutOrder");
}
#Order(2)
#Test
void addItemsInBasket() {
System.out.println("addItemsInBasket");
}
#Order(1)
#Test
void createUserAndLogin() {
System.out.println("createUserAndLogin");
}
}
Output :
createUserAndLogin
addItemsInBasket
checkoutOrder
By the way, specifying #TestMethodOrder(OrderAnnotation.class) looks like not needed (at least in the 5.4.0 version I tested).
Side note
About the question: is JUnit 5 the best choice to write integration tests? I don't think that it should be the first tool to consider (Cucumber and co may often bring more specific value and features for integration tests) but in some integration test cases, the JUnit framework is enough. So that is a good news that the feature exists.
The (as yet unreleased) change https://github.com/junit-team/junit/pull/386 introduces a #SortMethodsWith. https://github.com/junit-team/junit/pull/293 at least made the order predictable without that (in Java 7 it can be quite random).
Look at a JUnit report. JUnit is already organized by package. Each package has (or can have) TestSuite classes, each of which in turn run multiple TestCases. Each TestCase can have multiple test methods of the form public void test*(), each of which will actually become an instance of the TestCase class to which they belong. Each test method (TestCase instance) has a name and a pass/fail criteria.
What my management requires is the concept of individual TestStep items, each of which reports their own pass/fail criteria. Failure of any test step must not prevent the execution of subsequent test steps.
In the past, test developers in my position organized TestCase classes into packages that correspond to the part(s) of the product under test, created a TestCase class for each test, and made each test method a separate "step" in the test, complete with its own pass/fail criteria in the JUnit output. Each TestCase is a standalone "test", but the individual methods, or test "steps" within the TestCase, must occur in a specific order.
The TestCase methods were the steps of the TestCase, and test designers got a separate pass/fail criterion per test step. Now the test steps are jumbled, and the tests (of course) fail.
For example:
Class testStateChanges extends TestCase
public void testCreateObjectPlacesTheObjectInStateA()
public void testTransitionToStateBAndValidateStateB()
public void testTransitionToStateCAndValidateStateC()
public void testTryToDeleteObjectinStateCAndValidateObjectStillExists()
public void testTransitionToStateAAndValidateStateA()
public void testDeleteObjectInStateAAndObjectDoesNotExist()
public void cleanupIfAnythingWentWrong()
Each test method asserts and reports its own separate pass/fail criteria.
Collapsing this into "one big test method" for the sake of ordering loses the pass/fail criteria granularity of each "step" in the JUnit summary report. ...and that upsets my managers. They are currently demanding another alternative.
Can anyone explain how a JUnit with scrambled test method ordering would support separate pass/fail criteria of each sequential test step, as exemplified above and required by my management?
Regardless of the documentation, I see this as a serious regression in the JUnit framework that is making life difficult for lots of test developers.
Not sure I agree, If I want to test 'File Upload' and then test 'Data Inserted by File Upload' why would I not want these to be independent from each other? Perfectly reasonable I think to be able to run them separately rather than having both in a Goliath test case.
What you want is perfectly reasonable when test cases are being run as a suite.
Unfortunately no time to give a complete solution right now, but have a look at class:
org.junit.runners.Suite
Which allows you to call test cases (from any test class) in a specific order.
These might be used to create functional, integration or system tests.
This leaves your unit tests as they are without specific order (as recommended), whether you run them like that or not, and then re-use the tests as part of a bigger picture.
We re-use/inherit the same code for unit, integration and system tests, sometimes data driven, sometimes commit driven, and sometimes run as a suite.
JUnit 4 update
As of JUnit 4.13 #OrderWith, it is possible to reproduce the JUnit 5 #Order annotation. This solution better integrates with JUnit 4 than #RunWith a custom BlockJUnit4ClassRunner implementation.
Here's how I could replace method name ordering (#FixMethodOrder(MethodSorters.NAME_ASCENDING)) with an ordering by annotation.
#OrderWith(OrderAnnotation.class)
public class MyTest {
#Test
#Order(-1)
public void runBeforeNotAnnotatedTests() {}
#Test
public void notAnnotatedTestHasPriority0() {}
#Test
#Order(1)
public void thisTestHasPriority1() {}
#Test
#Order(2)
public void thisTestHasPriority2() {}
}
/**
* JUnit 4 equivalent of JUnit 5's {#code org.junit.jupiter.api.Order}
*/
#Retention(RetentionPolicy.RUNTIME)
#Target({ ElementType.METHOD })
public #interface Order {
/**
* Default order value for elements not explicitly annotated with {#code #Order}.
*
* #see Order#value
*/
int DEFAULT = 0;
/**
* The order value for the annotated element.
* <p>Elements are ordered based on priority where a lower value has greater
* priority than a higher value. For example, {#link Integer#MAX_VALUE} has
* the lowest priority.
*
* #see #DEFAULT
*/
int value();
}
import org.junit.runner.Description;
import org.junit.runner.manipulation.Ordering;
import org.junit.runner.manipulation.Sorter;
/**
* Order test methods by their {#link Order} annotation. The lower value has the highest priority.
* The tests that are not annotated get the default value {#link Order#DEFAULT}.
*/
public class OrderAnnotation extends Sorter implements Ordering.Factory {
public OrderAnnotation() {
super(COMPARATOR);
}
#Override
public Ordering create(Context context) {
return this;
}
private static final Comparator<Description> COMPARATOR = Comparator.comparingInt(
description -> Optional.ofNullable(description.getAnnotation(Order.class))
.map(Order::value)
.orElse(Order.DEFAULT));
}
The not annotated tests get a default priority of 0. The order of tests with the same priority is undetermined.
Gist: https://gist.github.com/jcarsique/df98e0bad9e88e8258c4ab34dad3c863
Inspired by:
Aman Goel's answer
Test execution order Wiki by JUnit team
JUnit 5 source code
See my solution here:
"Junit and java 7."
In this article I describe how to run junit tests in order - "just as in your source code".
Tests will be run, in order as your test methods appears in class file.
http://intellijava.blogspot.com/2012/05/junit-and-java-7.html
But as Pascal Thivent said, this is not a good practise.
As others have stated, tests should be ideally be independent of execution order. This makes tests less fragile, and allows them to be run independently (many IDEs allow you to select a test method and execute it independently of other tests).
That being said, for integration tests, some people prefer to rely on method ordering.
Starting with JUnit 4.13 you can define your own class to reorder tests by extending Ordering. See the JUnit wiki for more details. Here's an example using the built-in Alphanumeric class to order the tests alphanumerically using the test method name:
import org.junit.Test;
import org.junit.runner.OrderWith;
import org.junit.runner.manipulation.Alphanumeric;
#OrderWith(Alphanumeric.class)
public class TestMethodOrder {
#Test
public void testA() {
System.out.println("first");
}
#Test
public void testB() {
System.out.println("second");
}
#Test
public void testC() {
System.out.println("third");
}
}
For JUnit 4, putting this annotation on the test class solved the problem.
#FixMethodOrder(MethodSorters.JVM)
With JUnit 5.4, you can specify the order :
#Test
#Order(2)
public void sendEmailTestGmail() throws MessagingException {
you just need to annotate your class
#TestMethodOrder(OrderAnnotation.class)
https://junit.org/junit5/docs/current/user-guide/#writing-tests-test-execution-order
i'm using this in my project and it works very well !
You can use one of these piece of codes:
#FixMethodOrder(MethodSorters.JVM) OR #FixMethodOrder(MethodSorters.DEFAULT) OR #FixMethodOrder(MethodSorters.NAME_ASCENDING)
Before your test class like this:
#FixMethodOrder(MethodSorters.NAME_ASCENDING)
public class BookTest {...}
It's time to move to Junit5.
Here is a sample of what we could get:
#TestMethodOrder(MethodOrderer.OrderAnnotation.class)
class OrderedTests {
#Test
#Order(1)
void nullValues() {}
#Test
#Order(2)
void emptyValues() {}
#Test
#Order(3)
void validValues() {}
}
For Junit4, copy the logic that you have in several tests in one test method.
I've read a few answers and agree its not best practice, but the easiest way to order your tests - and the way that JUnit runs tests by default is by alphabetic name ascending.
So just name your tests in the alphabetic order that you want. Also note the test name must begin
with the word test. Just watch out for numbers
test12 will run before test2
so:
testA_MyFirstTest
testC_ThirdTest
testB_ATestThatRunsSecond
Please check out this one: https://github.com/TransparentMarket/junit. It runs the test in the order they are specified (defined within the compiled class file). Also it features a AllTests suite to run tests defined by sub package first. Using the AllTests implementation one can extend the solution in also filtering for properties (we used to use #Fast annotations but those were not published yet).
Here is an extension to JUnit that can produce the desired behavior: https://github.com/aafuks/aaf-junit
I know that this is against the authors of JUnit philosophy, but when using JUnit in environments that are not strict unit testing (as practiced in Java) this can be very helpful.
I ended up here thinking that my tests weren't run in order, but the truth is that the mess was in my async jobs. When working with concurrency you need to perform concurrency checks between your tests as well.
In my case, jobs and tests share a semaphore, so next tests hang until the running job releases the lock.
I know this is not fully related to this question, but maybe could help targeting the correct issue
If you want to run test methods in a specific order in JUnit 5, you can use the below code.
#TestMethodOrder(MethodOrderer.OrderAnnotation.class)
public class MyClassTest {
#Test
#Order(1)
public void test1() {}
#Test
#Order(2)
public void test2() {}
}

Start single Serenity scenario from command line

My team received ownership of a webapp. Tests are written with junit suites and serenity. Good things, there a good test coverage. Problem come when you need to run that single test/scenario that is still failing and you need to wait >30min to run everything.
How can I run a single scenario of this suite using mvn command line?
From code editor, it's hard to start single scenario as both suite and test classes contains important initialization code.
I've also tried argument '-Dtest=T1Test#T1Scenario1' without success.
Code snipplet:
#RunWith(Suite.class)
#Suite.SuiteClasses({
UserConfigASuite.class,
UserConfigBSuite.class,
UserConfigCSuite.class
})
public class AllTestSuite {
}
#RunWith(Suite.class)
#Suite.SuiteClasses({
T1Test.class,
T2Test.class,
//... Lots of other tests
})public class UserConfigASuite {
#BeforeClass
public static void beforeClass() {
//Required init code
}
#AfterClass
public static void afterClass() {
//Cleanup after test suite
}
}
#RunWith(SerenityRunner.class)
public class T1Test {
#Test
#Title("T1: scenario 1")
public void T1Scenario1() {
}
//... Lots of other scenarios
}
Just confirm first that your using supported surefire and junit version. For more details refer https://maven.apache.org/surefire/maven-surefire-plugin/examples/single-test.html
In case your using maven failsafe plugin then the syntax will vary little bit. Something like this
mvn -Dit.test=ITCircle#test* verify
Refer https://maven.apache.org/surefire/maven-failsafe-plugin/examples/single-test.html for more details.

TestNG parallel classesAndMethods without XML

I'm trying to have both classes and methods running in parallel.
For example:
Method Tests
public class MethodTests(){
#Test(groups = "testMe")
public void methodTestOne(){
...
}
#Test(groups = "testMe")
public void methodTestTwo(){
...
}
}
Class Tests
-> Hoping the Test annotation on the class level would do it
#Test
public class ClassTests(){
#Test(groups = "testMe")
public void classTestOne(){
...
}
#Test(groups = "testMe")
public void classTestTwo(){
...
}
}
I've included the surefire-plugin:
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-surefire-plugin</artifactId>
<version>2.19.1</version>
<configuration>
<parallel>all</parallel>
<threadCount>${threads}</threadCount>
...
</plugin>
Note: I've also tried classesAndMethods
I'm running from the command line like this:
mvn clean verify -Dgroups=testMe -Dthreads=3
I'm trying to accomplish the classTests run on one thread, and the method tests use a different thread after being complete.
Purpose:
In the beforeClass, I'm setting up the test, and then quickly asserting multiple things on a page, If one fails, I want it to still test the other things on the page.
Where as, if the method tests fail, I need it to completely stop.
I need both scenarios, and I do not want to use any XML.
I am not sure I quite understand the use case. But the only combinations of parallelism supported by TestNG are as below
tests - causes #Test methods inside tags to run in parallel.
instances - Causes #Test methods inside test class instances to run in parallel.
classes - causes test classes to run in parallel
methods - causes multiple #Test methods to run in parallel
I dont remember seeing something called all. I think that's applicable only to the JUnit provider for Maven surefire plugin. Please see here.
For your scenario as long as you have the #BeforeClass(alwaysRun=true) it would be executed for all groups and it would prevent the test methods that are dependent on the before class to be skipped.
The value that you should be using is classes, like this : <parallel>classes</parallel>

How to run test methods in specific order in JUnit4?

I want to execute test methods which are annotated by #Test in specific order.
For example:
public class MyTest {
#Test public void test1(){}
#Test public void test2(){}
}
I want to ensure to run test1() before test2() each time I run MyTest, but I couldn't find annotation like #Test(order=xx).
I think it's quite important feature for JUnit, if author of JUnit doesn't want the order feature, why?
I think it's quite important feature for JUnit, if author of JUnit doesn't want the order feature, why?
I'm not sure there is a clean way to do this with JUnit, to my knowledge JUnit assumes that all tests can be performed in an arbitrary order. From the FAQ:
How do I use a test fixture?
(...) The ordering of test-method invocations is not guaranteed, so testOneItemCollection() might be executed before testEmptyCollection(). (...)
Why is it so? Well, I believe that making tests order dependent is a practice that the authors don't want to promote. Tests should be independent, they shouldn't be coupled and violating this will make things harder to maintain, will break the ability to run tests individually (obviously), etc.
That being said, if you really want to go in this direction, consider using TestNG since it supports running tests methods in any arbitrary order natively (and things like specifying that methods depends on groups of methods). Cedric Beust explains how to do this in order of execution of tests in testng.
If you get rid of your existing instance of Junit, and download JUnit 4.11 or greater in the build path, the following code will execute the test methods in the order of their names, sorted in ascending order:
import org.junit.FixMethodOrder;
import org.junit.Test;
import org.junit.runners.MethodSorters;
#FixMethodOrder(MethodSorters.NAME_ASCENDING)
public class SampleTest {
#Test
public void testAcreate() {
System.out.println("first");
}
#Test
public void testBupdate() {
System.out.println("second");
}
#Test
public void testCdelete() {
System.out.println("third");
}
}
Migration to TestNG seems the best way, but I see no clear solution here for jUnit. Here is most readable solution / formatting I found for jUnit:
#FixMethodOrder( MethodSorters.NAME_ASCENDING ) // force name ordering
public class SampleTest {
#Test
void stage1_prepareAndTest(){};
#Test
void stage2_checkSomething(){};
#Test
void stage2_checkSomethingElse(){};
#Test
void stage3_thisDependsOnStage2(){};
#Test
void callTimeDoesntMatter(){}
}
This ensures stage2 methods are called after stage1 ones and before stage3 ones.
P.S. I feel this approach is better that jUnit 5.5 #Order annotation because it provides better notation for reader.
If the order is important, you should make the order yourself.
#Test public void test1() { ... }
#Test public void test2() { test1(); ... }
In particular, you should list some or all possible order permutations to test, if necessary.
For example,
void test1();
void test2();
void test3();
#Test
public void testOrder1() { test1(); test3(); }
#Test(expected = Exception.class)
public void testOrder2() { test2(); test3(); test1(); }
#Test(expected = NullPointerException.class)
public void testOrder3() { test3(); test1(); test2(); }
Or, a full test of all permutations:
#Test
public void testAllOrders() {
for (Object[] sample: permute(1, 2, 3)) {
for (Object index: sample) {
switch (((Integer) index).intValue()) {
case 1: test1(); break;
case 2: test2(); break;
case 3: test3(); break;
}
}
}
}
Here, permute() is a simple function which iterates all possible permuations into a Collection of array.
JUnit since 5.5 allows #TestMethodOrder(OrderAnnotation.class) on class and #Order(1) on test-methods.
JUnit old versions allow test methods run ordering using class annotations:
#FixMethodOrder(MethodSorters.NAME_ASCENDING)
#FixMethodOrder(MethodSorters.JVM)
#FixMethodOrder(MethodSorters.DEFAULT)
By default test methods are run in alphabetical order. So, to set specific methods order you can name them like:
a_TestWorkUnit_WithCertainState_ShouldDoSomething
b_TestWorkUnit_WithCertainState_ShouldDoSomething
c_TestWorkUnit_WithCertainState_ShouldDoSomething
Or
_1_TestWorkUnit_WithCertainState_ShouldDoSomething
_2_TestWorkUnit_WithCertainState_ShouldDoSomething
_3_TestWorkUnit_WithCertainState_ShouldDoSomething
You can find examples here.
Its one of the main issue which I faced when I worked on Junit and I came up with following solution which works fine for me:
import java.util.ArrayList;
import java.util.Collections;
import java.util.Comparator;
import java.util.List;
import org.junit.runners.BlockJUnit4ClassRunner;
import org.junit.runners.model.FrameworkMethod;
import org.junit.runners.model.InitializationError;
public class OrderedRunner extends BlockJUnit4ClassRunner {
public OrderedRunner(Class<?> clazz) throws InitializationError {
super(clazz);
}
#Override
protected List<FrameworkMethod> computeTestMethods() {
List<FrameworkMethod> list = super.computeTestMethods();
List<FrameworkMethod> copy = new ArrayList<FrameworkMethod>(list);
Collections.sort(copy, new Comparator<FrameworkMethod>() {
#Override
public int compare(FrameworkMethod f1, FrameworkMethod f2) {
Order o1 = f1.getAnnotation(Order.class);
Order o2 = f2.getAnnotation(Order.class);
if (o1 == null || o2 == null) {
return -1;
}
return o1.order() - o2.order();
}
});
return copy;
}
}
also create a interface like below:
#Retention(RetentionPolicy.RUNTIME)
#Target({ ElementType.METHOD})
public #interface Order {
public int order();
}
Now suppose you have class A where you have written several test cases like below:
(#runWith=OrderRunner.class)
Class A{
#Test
#Order(order = 1)
void method(){
//do something
}
}
So execution will start from method named "method()".
Thanks!
JUnit 5 update (and my opinion)
I think it's quite important feature for JUnit, if author of JUnit doesn't want the order feature, why?
By default, unit testing libraries don't try to execute tests in the order that occurs in the source file.
JUnit 5 as JUnit 4 work in that way. Why? Because if the order matters it means that some tests are coupled between them and that is undesirable for unit tests.
So the #Nested feature introduced by JUnit 5 follows the same default approach.
But for integration tests, the order of the test method may matter since a test method may change the state of the application in a way expected by another test method.
For example when you write an integration test for a e-shop checkout processing, the first test method to be executed is registering a client, the second is adding items in the basket and the last one is doing the checkout. If the test runner doesn't respect that order, the test scenario is flawed and will fail.
So in JUnit 5 (from the 5.4 version) you have all the same the possibility to set the execution order by annotating the test class with #TestMethodOrder(OrderAnnotation.class) and by specifying the order with #Order(numericOrderValue) for the methods which the order matters.
For example :
#TestMethodOrder(OrderAnnotation.class)
class FooTest {
#Order(3)
#Test
void checkoutOrder() {
System.out.println("checkoutOrder");
}
#Order(2)
#Test
void addItemsInBasket() {
System.out.println("addItemsInBasket");
}
#Order(1)
#Test
void createUserAndLogin() {
System.out.println("createUserAndLogin");
}
}
Output :
createUserAndLogin
addItemsInBasket
checkoutOrder
By the way, specifying #TestMethodOrder(OrderAnnotation.class) looks like not needed (at least in the 5.4.0 version I tested).
Side note
About the question: is JUnit 5 the best choice to write integration tests? I don't think that it should be the first tool to consider (Cucumber and co may often bring more specific value and features for integration tests) but in some integration test cases, the JUnit framework is enough. So that is a good news that the feature exists.
The (as yet unreleased) change https://github.com/junit-team/junit/pull/386 introduces a #SortMethodsWith. https://github.com/junit-team/junit/pull/293 at least made the order predictable without that (in Java 7 it can be quite random).
Look at a JUnit report. JUnit is already organized by package. Each package has (or can have) TestSuite classes, each of which in turn run multiple TestCases. Each TestCase can have multiple test methods of the form public void test*(), each of which will actually become an instance of the TestCase class to which they belong. Each test method (TestCase instance) has a name and a pass/fail criteria.
What my management requires is the concept of individual TestStep items, each of which reports their own pass/fail criteria. Failure of any test step must not prevent the execution of subsequent test steps.
In the past, test developers in my position organized TestCase classes into packages that correspond to the part(s) of the product under test, created a TestCase class for each test, and made each test method a separate "step" in the test, complete with its own pass/fail criteria in the JUnit output. Each TestCase is a standalone "test", but the individual methods, or test "steps" within the TestCase, must occur in a specific order.
The TestCase methods were the steps of the TestCase, and test designers got a separate pass/fail criterion per test step. Now the test steps are jumbled, and the tests (of course) fail.
For example:
Class testStateChanges extends TestCase
public void testCreateObjectPlacesTheObjectInStateA()
public void testTransitionToStateBAndValidateStateB()
public void testTransitionToStateCAndValidateStateC()
public void testTryToDeleteObjectinStateCAndValidateObjectStillExists()
public void testTransitionToStateAAndValidateStateA()
public void testDeleteObjectInStateAAndObjectDoesNotExist()
public void cleanupIfAnythingWentWrong()
Each test method asserts and reports its own separate pass/fail criteria.
Collapsing this into "one big test method" for the sake of ordering loses the pass/fail criteria granularity of each "step" in the JUnit summary report. ...and that upsets my managers. They are currently demanding another alternative.
Can anyone explain how a JUnit with scrambled test method ordering would support separate pass/fail criteria of each sequential test step, as exemplified above and required by my management?
Regardless of the documentation, I see this as a serious regression in the JUnit framework that is making life difficult for lots of test developers.
Not sure I agree, If I want to test 'File Upload' and then test 'Data Inserted by File Upload' why would I not want these to be independent from each other? Perfectly reasonable I think to be able to run them separately rather than having both in a Goliath test case.
What you want is perfectly reasonable when test cases are being run as a suite.
Unfortunately no time to give a complete solution right now, but have a look at class:
org.junit.runners.Suite
Which allows you to call test cases (from any test class) in a specific order.
These might be used to create functional, integration or system tests.
This leaves your unit tests as they are without specific order (as recommended), whether you run them like that or not, and then re-use the tests as part of a bigger picture.
We re-use/inherit the same code for unit, integration and system tests, sometimes data driven, sometimes commit driven, and sometimes run as a suite.
JUnit 4 update
As of JUnit 4.13 #OrderWith, it is possible to reproduce the JUnit 5 #Order annotation. This solution better integrates with JUnit 4 than #RunWith a custom BlockJUnit4ClassRunner implementation.
Here's how I could replace method name ordering (#FixMethodOrder(MethodSorters.NAME_ASCENDING)) with an ordering by annotation.
#OrderWith(OrderAnnotation.class)
public class MyTest {
#Test
#Order(-1)
public void runBeforeNotAnnotatedTests() {}
#Test
public void notAnnotatedTestHasPriority0() {}
#Test
#Order(1)
public void thisTestHasPriority1() {}
#Test
#Order(2)
public void thisTestHasPriority2() {}
}
/**
* JUnit 4 equivalent of JUnit 5's {#code org.junit.jupiter.api.Order}
*/
#Retention(RetentionPolicy.RUNTIME)
#Target({ ElementType.METHOD })
public #interface Order {
/**
* Default order value for elements not explicitly annotated with {#code #Order}.
*
* #see Order#value
*/
int DEFAULT = 0;
/**
* The order value for the annotated element.
* <p>Elements are ordered based on priority where a lower value has greater
* priority than a higher value. For example, {#link Integer#MAX_VALUE} has
* the lowest priority.
*
* #see #DEFAULT
*/
int value();
}
import org.junit.runner.Description;
import org.junit.runner.manipulation.Ordering;
import org.junit.runner.manipulation.Sorter;
/**
* Order test methods by their {#link Order} annotation. The lower value has the highest priority.
* The tests that are not annotated get the default value {#link Order#DEFAULT}.
*/
public class OrderAnnotation extends Sorter implements Ordering.Factory {
public OrderAnnotation() {
super(COMPARATOR);
}
#Override
public Ordering create(Context context) {
return this;
}
private static final Comparator<Description> COMPARATOR = Comparator.comparingInt(
description -> Optional.ofNullable(description.getAnnotation(Order.class))
.map(Order::value)
.orElse(Order.DEFAULT));
}
The not annotated tests get a default priority of 0. The order of tests with the same priority is undetermined.
Gist: https://gist.github.com/jcarsique/df98e0bad9e88e8258c4ab34dad3c863
Inspired by:
Aman Goel's answer
Test execution order Wiki by JUnit team
JUnit 5 source code
See my solution here:
"Junit and java 7."
In this article I describe how to run junit tests in order - "just as in your source code".
Tests will be run, in order as your test methods appears in class file.
http://intellijava.blogspot.com/2012/05/junit-and-java-7.html
But as Pascal Thivent said, this is not a good practise.
As others have stated, tests should be ideally be independent of execution order. This makes tests less fragile, and allows them to be run independently (many IDEs allow you to select a test method and execute it independently of other tests).
That being said, for integration tests, some people prefer to rely on method ordering.
Starting with JUnit 4.13 you can define your own class to reorder tests by extending Ordering. See the JUnit wiki for more details. Here's an example using the built-in Alphanumeric class to order the tests alphanumerically using the test method name:
import org.junit.Test;
import org.junit.runner.OrderWith;
import org.junit.runner.manipulation.Alphanumeric;
#OrderWith(Alphanumeric.class)
public class TestMethodOrder {
#Test
public void testA() {
System.out.println("first");
}
#Test
public void testB() {
System.out.println("second");
}
#Test
public void testC() {
System.out.println("third");
}
}
For JUnit 4, putting this annotation on the test class solved the problem.
#FixMethodOrder(MethodSorters.JVM)
With JUnit 5.4, you can specify the order :
#Test
#Order(2)
public void sendEmailTestGmail() throws MessagingException {
you just need to annotate your class
#TestMethodOrder(OrderAnnotation.class)
https://junit.org/junit5/docs/current/user-guide/#writing-tests-test-execution-order
i'm using this in my project and it works very well !
You can use one of these piece of codes:
#FixMethodOrder(MethodSorters.JVM) OR #FixMethodOrder(MethodSorters.DEFAULT) OR #FixMethodOrder(MethodSorters.NAME_ASCENDING)
Before your test class like this:
#FixMethodOrder(MethodSorters.NAME_ASCENDING)
public class BookTest {...}
It's time to move to Junit5.
Here is a sample of what we could get:
#TestMethodOrder(MethodOrderer.OrderAnnotation.class)
class OrderedTests {
#Test
#Order(1)
void nullValues() {}
#Test
#Order(2)
void emptyValues() {}
#Test
#Order(3)
void validValues() {}
}
For Junit4, copy the logic that you have in several tests in one test method.
I've read a few answers and agree its not best practice, but the easiest way to order your tests - and the way that JUnit runs tests by default is by alphabetic name ascending.
So just name your tests in the alphabetic order that you want. Also note the test name must begin
with the word test. Just watch out for numbers
test12 will run before test2
so:
testA_MyFirstTest
testC_ThirdTest
testB_ATestThatRunsSecond
Please check out this one: https://github.com/TransparentMarket/junit. It runs the test in the order they are specified (defined within the compiled class file). Also it features a AllTests suite to run tests defined by sub package first. Using the AllTests implementation one can extend the solution in also filtering for properties (we used to use #Fast annotations but those were not published yet).
Here is an extension to JUnit that can produce the desired behavior: https://github.com/aafuks/aaf-junit
I know that this is against the authors of JUnit philosophy, but when using JUnit in environments that are not strict unit testing (as practiced in Java) this can be very helpful.
I ended up here thinking that my tests weren't run in order, but the truth is that the mess was in my async jobs. When working with concurrency you need to perform concurrency checks between your tests as well.
In my case, jobs and tests share a semaphore, so next tests hang until the running job releases the lock.
I know this is not fully related to this question, but maybe could help targeting the correct issue
If you want to run test methods in a specific order in JUnit 5, you can use the below code.
#TestMethodOrder(MethodOrderer.OrderAnnotation.class)
public class MyClassTest {
#Test
#Order(1)
public void test1() {}
#Test
#Order(2)
public void test2() {}
}

Javolution test patterns, dos and don'ts

What are the patterns and dos and don'ts when one is writing tests for Javolution tests? In particular I was wondering:
TestCase.execute() does not allow throwing of exceptions. How to deal with them? Rethrow as RuntimeException or store in a variable and assert in TestCase.validate() or something?
Are there any graphical runners that show you the tests that fail, i.e. in Eclipse? Perhaps someone wrote a JUnit-Wrapper such that I could use the Eclipse JUnit Runner?
The javadoc and javolution sources give some examples and design rationale.
See also an article on serverside.
Javolution tests contain exactly one test, and the exercising of the tested code is separated from the validation into different methods execute() and validate(). Thus the same testclass can be used both for regression tests and speed tests (where the call to validate() is omitted). Also the execution of many tests is trivially parallelizable.
A disadvantages of this separation is: you will get more memory consumption, since the results of the execution of the exercised code needs to be saved until calling validate(). (Freeing those results in tearDown is probably a good idea.)
And if validate comes from a different class than exercise then it might be difficult to debug a failure.
You can get some kind of graphical testrunner by using the following JUnit adapter and running it in eclipse. You can start / debug the failed tests separately. Unfortunately the graphical representation does not include anything about the actual test - it just shows the numbers [0], [1], etc.
#RunWith(Parameterized.class)
public class JavolutionJUnit4Adapter {
protected final javolution.testing.TestCase test;
public JavolutionJUnit4Adapter(javolution.testing.TestCase testcase) {
this.test = testcase;
}
#org.junit.Test
public void executeTest() throws Exception {
enter(REGRESSION);
try {
new javolution.testing.TestSuite() {
#Override
public void run() {
test(test);
}
}.run();
} finally {
exit();
}
}
#Parameters
public static Collection<javolution.testing.TestCase[]> data() {
javolution.testing.TestSuite fp = new WhateverSuiteYouWantToRun();
List<javolution.testing.TestCase> tests = fp.getTestCases();
Collection<javolution.testing.TestCase[]> res = new ArrayList<javolution.testing.TestCase[]>();
for (javolution.testing.TestCase t : tests) {
res.add(new javolution.testing.TestCase[] { t });
}
return res;
}
}

Categories