Also v2.6.2 breaks existing test (Assert.Equal on Dictionary with null values)

[gioce90]

I have a similar problem to #2821.
Every version of xunit.assert and xunit > 2.5.3 give me an error, also with the 2.6.2.

I tested with version 2.6.2 (of both xunit.assert and xunit), and I think there is still a problem with collections like Dictionary when there is a Value field with null value.

This is my first test, that is totally fine:

        [Fact]
        public void MyTest__OKAY()
        {
            IDictionary<string, string> dict1 = new Dictionary<string, string>
            {
                { "A", "The" },
                { "B", "empty" },
                { "C", "" },
                { "D", "test" },
                { "E", "is" },
                { "F", "OKAY" },
            };
            IDictionary<string, string> dict2 = new Dictionary<string, string>
            {
                { "A", "The" },
                { "B", "empty" },
                { "C", "" },
                { "D", "test" },
                { "E", "is" },
                { "F", "OKAY" },
            };
            Assert.Equal(dict1, dict2);
            Assert.Equal(dict2, dict1);
        }

This second one instead, where i just put a null value, fails:

        [Fact]
        public void MyTest__WRONG()
        {
            IDictionary<string, string> dict1 = new Dictionary<string, string>
            {
                { "A", "The" },
                { "B", "null" },
                { "C", null },
                { "D", "test" },
                { "E", "is" },
                { "F", "NOT" },
            };
            IDictionary<string, string> dict2 = new Dictionary<string, string>
            {
                { "A", "The" },
                { "B", "null" },
                { "C", null },
                { "D", "test" },
                { "E", "is" },
                { "F", "NOT" },
            };
            Assert.Equal(dict2, dict1);
            Assert.Equal(dict1, dict2);
        }

But if I go back to version "2.5.3", both tests are fine. I think I've found another regression here.

[ladeak]

Duplicate of #2824