Inconsistent Tuple behavior.

[constructor-igor]

Next test fails (NUnit 2.6.4 and 3.0 Beta 4)

        [Test]
        public void TupleIntVsDouble()
        {
            Tuple<int, int> actual = Tuple.Create(1, 2);
            Tuple<double, double> expected = Tuple.Create(1.0, 2.0 );

            Assert.That(actual, Is.EqualTo(expected));
        }

with unclear message (R#):

VS adapter:

But "same" tests pass:

        [Test]
        public void IntVsDouble()
        {
            int actual = 1;
            double expected = 1.0;

            Assert.That(actual, Is.EqualTo(expected));
        }
        [Test]
        public void ListIntVsDouble()
        {
            List<int>actual = new List<int>{1, 2};
            List<double> expected = new List<double>{1.0, 2.0};

            Assert.That(actual, Is.EquivalentTo(expected));
        }

[rprouse]

I've moved this out to 3.2 because comparing two Tuples of different types isn't valid C#. In other words, Assert.That(actual == expected) will not compile.

It highlights a sort of problem with NUnit in that we try to be much more lenient with equality so people expect the same amount of leniency with all types.

I think that Tuples are a valid extension of our leniency, but when we work on this issue, I would like to see us refactor NUnitEqualityComparer. It is becoming a monster of a class. I would prefer to see it as the central coordinator of a series of equality types that each handles specific equality conversions.

[CharliePoole]

@rprouse Good approach. Let's do that for 3.2.

On Wed, Oct 14, 2015 at 6:47 AM, Rob Prouse notifications@github.com
wrote:

I've moved this out to 3.2 because comparing two Tuples of different types
isn't valid C#. In other words, Assert.That(actual == expected) will not
compile.

It highlights a sort of problem with NUnit in that we try to be much more
lenient with equality so people expect the same amount of leniency with all
types.

I think that Tuples are a valid extension of our leniency, but when we
work on this issue, I would like to see us refactor NUnitEqualityComparer.
It is becoming a monster of a class. I would prefer to see it as the
central coordinator of a series of equality types that each handles
specific equality conversions.

—
Reply to this email directly or view it on GitHub
#893 (comment).

[CharliePoole]

I removed easyfix because refactoring NUnitEqualityComparer is not something I'd like a first-time contributor to try. Granted, @rprouse , it's easy for you or me... :-)

[jnm2]

@nunit/framework-team We just had the same discussion on the ValueTuple type now that C# has that type backing its tuple language feature. Would we be in agreement that the same purposes apply to Tuple? I was under the impression that Tuple plays a similar role in F#'s tuple language feature that ValueTuple now does in C#.

@mikkelbu Would you have any interest in this?
I'm thinking it's going to be almost identical to #2290. If not, we can put it up for grabs since the implementation spec would be #2290. 😄

[mikkelbu]

@jnm2 Sure. I can take a look at it.