`Math.Min(double/float)` do not handle NaN correctly

[CoryCharlton]

Library/API/IoT binding

System.Math

Visual Studio version

No response

.NET nanoFramework extension version

No response

Target name(s)

No response

Firmware version

No response

Device capabilities

No response

Description

According to IEEE 754:2019 (based on comments in the .NET Core implementation) NaN should be returned if either val1 or val2 are NaN.

There is also an opportunity here to optimize the Max and Min performance by removing one of the IsNan checks as demonstrated in the .NET Core implementation.

I'll look into implementing this change.

How to reproduce

Run these tests:

    [TestClass]
    public class System_Math_Tests
    {
        [TestMethod]
        public void Max_Double_returns_NaN_if_val1_is_NaN()
        {
            Assert.IsTrue(double.IsNaN(System.Math.Max(double.NaN, Math.PI)));
        }

        [TestMethod]
        public void Max_Double_returns_NaN_if_val2_is_NaN()
        {
            Assert.IsTrue(double.IsNaN(System.Math.Max(Math.PI, double.NaN)));
        }

        [TestMethod]
        public void Max_Float_returns_NaN_if_val1_is_NaN()
        {
            Assert.IsTrue(double.IsNaN(System.Math.Max(float.NaN, (float) Math.PI)));
        }

        [TestMethod]
        public void Max_Float_returns_NaN_if_val2_is_NaN()
        {
            Assert.IsTrue(double.IsNaN(System.Math.Max((float) Math.PI, float.NaN)));
        }

        [TestMethod]
        public void Min_Double_returns_NaN_if_val1_is_NaN()
        {
            // This test incorrectly returns PI
            Assert.IsTrue(double.IsNaN(System.Math.Min(double.NaN, Math.PI)));
        }

        [TestMethod]
        public void Min_Double_returns_NaN_if_val2_is_NaN()
        {
            Assert.IsTrue(double.IsNaN(System.Math.Min(Math.PI, double.NaN)));
        }

        [TestMethod]
        public void Min_Float_returns_NaN_if_val1_is_NaN()
        {
            // This test incorrectly returns PI
            Assert.IsTrue(double.IsNaN(System.Math.Min(float.NaN, (float) Math.PI)));
        }

        [TestMethod]
        public void Min_Float_returns_NaN_if_val2_is_NaN()
        {
            Assert.IsTrue(double.IsNaN(System.Math.Min((float) Math.PI, float.NaN)));
        }
    }

Expected behaviour

NaN is returned when val1, val2, or both val1 and val2 are NaN.

Screenshots

No response

Sample project or code

No response

Aditional information

No response