What is the correct result of Equals<NaN, NaN>?

[SparkieLabs]

Under IEEE rules NaN is not equal to NaN.

The way this is implemented for double is that the == operator obeys the IEEE rule but Equals doesn't:

    [Fact]
    public void DoubleEqualityOpNaNTest()
    {
        double a = double.NaN;
        // Test passes
        Assert.False(a == a);
    }

    [Fact]
    public void DoubleEqualsNaNTest()
    {
        double a = double.NaN;
        // Test fails
        Assert.False(a.Equals(a));
    }

Which is strange as I would expect == and Equals to be consistent, but it is even mentioned in the docs.

The non-generic Vector types return false for both the operator and Equals method, which is what is currently implemented here.

Also note that if you use the equality operator generated for records with double properties, as it uses the double Equals method, neither the == operator nor Equals method returns false:

    readonly record struct DoublePropertyRecord(double Value);

    [Fact]
    public void DoublePropertyRecordEqualsNaNTest1()
    {
        DoublePropertyRecord a = new DoublePropertyRecord(double.NaN);
        // Test fails
        Assert.False(a == a);
    }

    [Fact]
    public void DoublePropertyRecordEqualsNaNTest2()
    {
        DoublePropertyRecord a = new DoublePropertyRecord(double.NaN);
        // Test fails
        Assert.False(a.Equals(a));
    }

[SparkieLabs]

Discussed here System.Numerics.Vector2.Equals doesn't use float.Equals in Equals implementation