assertEqual does not support nullptr

[hlovdal]

Trying to test that a function returns a null pointer, e.g.

	assertEqual(nullptr, result);

does not work

.../ArduinoUnitTests.h:69:56: error: ambiguous overload for ‘operator<<’ (operand types are ‘std::basic_ostream<char>’ and ‘std::nullptr_t’)
               cerr << "      " << lhsRelevance << ": " << lhs << endl;

I tried to fix it by extracting that line into a template function and then make a template specialisation for nullptr but did not get the compiler to select it, and have put that on ice for now since there is a simple workaround, just adding

#define nullptr (void *)NULL

at the top of the test file. But it would be nice to eventually have this fixed.

[ianfixes]

Yup, this seems like part of C++11 that needs to be added to the mocks. What's the data type of result in the code you posted? I'll add that to a unit test.

[ianfixes]

I will add nullptr and assertNull(result)

[hlovdal]

Those updates still do not fix the issue with assertEqual. The problem is that onAssert tries to print the argument of type nullptr which it is not able to. There needs to be an additional unit test:

unittest(nullpointer_equal)
{
  int* myPointer = NULL;
  int **notNullPointer = &myPointer;
  assertEqual(nullptr, myPointer);
  assertNotEqual(nullptr, notNullPointer);

#if 0
  // The following should not compile, which they currently do not but in a
  // cryptic way. If possible and with low effort an explicit denial which
  // produces an explanatory error would be nice.
  assertLessOrEqual(nullptr, myPointer);
  assertMoreOrEqual(nullptr, notNullPointer);
  assertLess(nullptr, myPointer);
  assertMore(nullptr, notNullPointer);
  assertFalse(nullptr);
  assertTrue(nullptr);
  assertAnyLhsRhsFunction(..., nullptr);
#endif
}

[ianfixes]

Thanks, those unit tests are exactly what was needed.

[ianfixes]

@hlovdal You gave me an excellent idea here.

I've set up a new branch called tdd (https://github.com/ianfixes/arduino_ci/tree/tdd). If you can write unit tests that show a problem, make a PR using the tdd branch as a base. That will give me something to work from when I go about fixing the issue, and not require me to break the master branch in the process.

[ianfixes]

This fix is live in the latest version, sorry for the arbitrary delay there