MS VS solution structure w.r.t. dependency injection unit tests?

Question

Is the following MS VS solution structure and classes OK w.r.t. easy unit tests with dependency injection? Would your unit tests look similar?

If you call the calculations below over a WPF module, would you skip a test over the WPF since complicated, or would you still conisder testing it over the WPF module ?

Solution

1. ProjectWork

   Classes

   • IDev.cs

     public interface IDev
     {
      IToMeasure Measure();
     }
     public interface IToMeasure
     {
      double Value{ get; set; }
     }
     

   • CalculatingUnit.cs

     public class CalculatingUnit
     {
      int count = 0,
      maxMeasuredValues = 20;
      public double doMeasures(IDev myDevice)
      {
      // simulating measuring each Xms
      while (count < maxMeasuredValues ) 
      {
      return myDevice.Measure();
      count++;
      }
      }
     }
     

   • MyDevice.cs

     // non-deterministic (random) values
     public class MyDevice: IDev
     {
      private readonly MyRandom myRandomVar = new Random(0); 
      public IToMeasure Measure()
      {
      var meas = new Measure { Value = myRandomVar.NextRandom(2, 0.5) };
      return meas;
      }
     }
     

   • TestDevice.cs

     // deterministic values
     public class TestDevice : IDevice
     {
      int nrValues = 6;
      int gIdx = 0,
      gCount = 0;
      int[] testMeasures = { 1, 2, 3, 4, 5, 6 };
      public IToMeasure Measure()
      {
      gIdx = gCount % nrValues;
      var meas = new Measure { Value = testMeasures[gIdx] };
      gCount++;
      return meas;
      }
     }
     

   • myDeviceWPF.cs // it seems complicated testing over WPF, so I dont test with it, would you also skip it ?

     class myDeviceViewModel : INotifyPropertyChanged
     {
      ...
     }
     

2. ProjectTest

   • myDevice_Test.cs

     [TestClass]
     public class myDevice_Test
     {
      [TestMethod]
      public void Test_doMeasures()
      {
      double actual = 1;
      double expected = 21.068;
      TestDevice myDevice;
      CalculatingUnit myCalculatingUnit;
      myDevice = new TestDevice();
      myCalculatingUnit = new CalculatingUnit();
      actual = myCalculatingUnit.doMeasures(myDevice );
      Assert.AreEqual(expected, actual , 0.001, "Wrong calculations");
      }
     }
     

• Malachi: in your MyDevice.cs file you are missing a closing } sure it was a typo but thought I would point it out
• HeinrichStack: @Malachi : 10x. Yes, it was a typo. I just corrected it.

Answer 1

First, some quality/stylistic comments:

This won't do what you want/isn't legal:

 while (count < maxMeasuredValues ) 
 {
 return myDevice.Measure();
 count++;
 }
// non-deterministic (random) values
private readonly MyRandom myRandomVar = new Random(0); 

Random initialized with a fixed seed is deterministic. Even with no seed (current time) it is technically deterministic, but that's pickier. Also, I don't see how you are using MyDevice in the code sample.

 double actual = 1;
 double expected = 21.068;
 TestDevice myDevice;
 CalculatingUnit myCalculatingUnit;
 myDevice = new TestDevice();
 myCalculatingUnit = new CalculatingUnit();
 actual = myCalculatingUnit.doMeasures(myDevice );

In general, prefer to declare and assign a value at the same time. This will make the functions shorter and more readable.

Other than that, the organization looks fine.