Develop Reference

C# MSTest Test Isolation

I am using MSTest to write integration tests which call a DLL. I need to load the DLL with different values based on the test, but the values in the DLL are persisting after a test is run.
Is there a way to isolate each test so that they each run in their own process or stack?
added code:
[TestClass]
public class TestClass1
{
[TestMethod]
public void Test1()
{
LoadDLL(1);
Assert.AreEqual(1, ReadDLL());
}
[TestMethod]
public void Test2()
{
Assert.AreEqual(1, ReadDLL());
}
}
If Test1 runs first, Test2 passes even though LoadDLL() should have needed to be called. Conversely, if Test2 runs first, it will fail.
Ideally LoadDLL would go in a [TestInitialize()] block.

#usr1 It seems appdomain no longer exists in .nrt core, you have to use assembly load context. Try this and then try to remove the context http://www.michael-whelan.net/replacing-appdomain-in-dotnet-core/

Nunit not calling setup once per text fixture

I'm new to Nunit and am trying to run 2 Test Fixtures, A & B. Within each Fixture I have a unique Setup method for each. However, when I click "Run All" in the "Test Explorer" in Visual Studio, the test setup for Fixture A is called (it was executed first) and Setup for Fixture B is ignored. I also get the same behavior when running all tests via command line. Below is my code:
Fixture A
[TestFixture]
public class A
{
[SetUp]
public void SetupTest()
{
// ...Setup for Fixture A
}
[Test, Order(1)]
public void TestForFixtureA()
{
// ...perform test
}
}
Fixture B
[TestFixture]
public class B
{
[SetUp]
public void SetupTest()
{
// ...Setup for Fixture B
}
[Test]
public void TestForFixtureB()
{
// ...perform test
}
}
What is the correct way to get Setup methods to execute per Fixture?

You are using the incorrect attribute for setup at the test fixture level. The attribute you should be using is [SetUpFixture]. Information about this can be found in the nunit documentation
Here is a list of all the setup attributes and their usages taken from the documentation:
SetUpAttribute is now used exclusively for per-test setup.
TearDownAttribute is now used exclusively for per-test teardown.
OneTimeSetUpAttribute is used for one-time setup per test-run. If you
run n tests, this event will only occur once.
OneTimeTearDownAttribute is used for one-time teardown per test-run.
If you run n tests, this event will only occur once
SetUpFixtureAttribute continues to be used as at before, but with
changed method attributes.
This doesn't seem to explain the bizzare behaviour you are seeing, as setup should be ran per-test, but using the correct attributes couldn't hurt.

If you intend your setup to be run once per fixture, use [OneTimeSetUp]. But if you intend it to run once per test within the fixture, then [SetUp] is correct. We can't tell what you intend from the code.
Whichever one you use, the setups should all run. The only situation in which [OneTimeSetUp] will run but [SetUp] will not is when no individual tests are found within the fixture. Are all the tests being recognized?
I suggest you verify very clearly that the setup is not being run. The easiest way is to temporarily create some output from the setup method.

Prevent NUnit tests to run depending on environment

I'm struggle with this one quite a long time now. Some background: I created my automated test framework using Selenium. With one project being pure NUnit tests and second one that does all the work for the tests. Now, in test project I keep the directories to all environments I run my tests against. So far many of my tests were read-only and didn't bother much about if tests did not run on environment that they supposed to. This has changed as I started to run some 'write' tests.
I need to prevent this 'Write' tests to run on any other environment then localhost. So far I tried to use method attributes and getting test method names on run time and do work then but this is not quite efficient. Can you guys point me a good solution? Thanks!

I would tag the tests to exclude with a particular category name then define a SetUp function which will stop the tests from running if they are tagged with that name and if you are on a particular environment such as Production. Place the SetUp function in a BaseClass and have all your test fixtures inherit it. The SetUp function will run before every test and prevent it from running if it needs to.
Something like this:
public class BaseSetup
{
protected const string CategoryToExclude = "Write";
[SetUp]
public void Init()
{
string env = ConfigurationManager.GetEnvironment();
if ( env == Constants.Environments.PROD && (TestContext.CurrentContext.Test.Properties["Categories"].Contains(CategoryToExclude)))
{
Assert.Inconclusive(String.Format("Cannot run this test on environment: {0}", env));
}
}
}
[TestFixture]
public class UnitTests : BaseSetup
{
[Test]
[Category(CategoryToExclude)]
public void TestMethod1()
{
Console.WriteLine("TestMethod1");
}
[Test]
public void TestMethod2()
{
Console.WriteLine("TestMethod2");
}
}
Hope this helps!

NUnit have category attribute.
You can group tests by categories and run only wanted categories.

writing a test using moq unit testing framework

I'm a bit confused with writing unit tests in C#. I've written the following class for learning Moq. I've noticed the [SetUp] is actually a reference from NUnit. How can I write a test class that only uses one framework or another or whether that's possible? If I want to use Moq what attributes am I missing to successfully run this test? I know there's [TestFixture], [TestMethod] etc.. but which one do I use for Moq!
Thanks,
James
public class CatalogCommandTests
{
private Mock<IReferenceData> _mockReferenceData;
[SetUp]
public void TestInitialize()
{
_mockReferenceData = new Mock<IReferenceData>();
}
public void TestMyGetMethodReturnsListOfStrings()
{
var contractsList = new List<Contract>();
_mockReferenceData.Setup(x => x.MyGetMethod()).Returns(contractsList);
}
}

Your mock looks good. To get the mock of IReferenceData you have to call _mockReferenceData.Object.
A method decorated with the SetUp Attribute will be executed before each test method is called.
(See: Setup).
If you want that TestMyGetMethodReturnsListOfStrings gets called you have to decorate the method with the Test attribute
(See: Test).

NUnit test cases not run from inherited class

I have a base test class containing some test cases and some ordinary tests:
[TestFixture]
public abstract class TestBase
{
[TestCase(1)]
[TestCase(2)]
[TestCase(3)]
public void TestA(int value)
{
// Perform test
}
[Test]
public void TestB()
{
// Perform test
}
}
These tests are run from child classes which set up the environment in different ways. The child classes only contains setup methods, no tests.
[TestFixture]
public class LocalDatabaseTest : TestBase
{
[SetUp]
public void SetUp()
{
// Set up environment to use local db
}
}
I'm using ReSharper 6.1.1000.82 to run all tests in LocalDatabaseTest, but only the ordinary tests are run. The tests using TestCase does not get any result. If I select Run All on TestA in the TestBase class, all test cases are run (including the other child classes). I'm using NUnit 2.6.2.12296. Any ideas on what I've done wrong?

You've done nothing wrong.
If you open your test dll via NUnit test runner you will see all test are running successfully.
(I just verified your code with NUnit 2.6.2).
Regarding the reason of ignoring parameterized tests on Resharper: It seems there is some issue with Resharper test runner which causes such behavior.
So, my suggestion is to use NUnit to run parameterized tests.
Btw, Resharper 7 has better support NUnit parameterized tests. And probably this issue won't appear in the latest Resharper version.