Random.Next returns negative numbers

[MatthiasJentsch]

Details about Problem

nanoFramework area: nanoCLR preview 791 / nanoFramework.CoreLibrary preview 62

Worked before? No

Detailed repro steps so we can see the same problem

The Random.Next returns not only values >= 0. I think that's caused by the nanoFramework.CoreLibrary Random.Next implementation. There is the uint32 return value from the unmanged part returned as int32

In nanoFramework.CoreLibrary:

/// <summary>
/// Returns a nonnegative random number.
/// </summary>
/// <returns>A 32-bit signed integer greater than or equal to zero and
/// less than MaxValue.</returns>
[MethodImpl(MethodImplOptions.InternalCall)]
public virtual extern int Next();

But in nanoCLR the return type is uint32_t

[josesimoes]

You've spotted an interesting situation! 😄
The API doc states that's an Int32 with max of Int32.MaxValue.

The lower layer functions (on all targets) return an uint32_t which is happily loaded on the return var with calls to stack.SetResult_I4( ... );. If the random value happens to have it's MSB set it will turn into a negative value.

Without putting too much thinking into this, it seems that we can 1) cast, 2) request a new one until it's in range, 3) change the code on the lower layer to keep it "in range" or none of the previous... 😉

[doingnz]

I would vote for 3 and suggest the following pseudo-code (value a place holder for the existing variable that is uint32_t)

if (value > (uint32_t)Int32.MaxValue) {value &= (uint32_t)Int32.MaxValue;} 

Ensures value is less than or equal to Int32.MaxValue, and the casts ensure no compiler warnings about comparison of signed/unsigned variables.