What causes "min must not exceed max (Parameter 'max')" in dotnet/machinelearning?

Verify the min/max computation: ensure max is always >= min before the call, e.g. use Math.Max(min, currentSize) for the max parameter.

dotnet/machinelearning · error · ArgumentOutOfRangeException

min must not exceed max (Parameter 'max')

Error message

min must not exceed max (Parameter 'max')

What it means

ArrayUtils.EnsureSize<T> grows (or optionally shrinks) a backing array so its length is at least 'min' and at most 'max'. The API contract requires min <= max; when min exceeds max the method throws ArgumentOutOfRangeException on parameter 'max' before doing any resizing. This invariant keeps callers from requesting contradictory bounds that the resize logic cannot satisfy.

Source

Thrown at src/Microsoft.ML.Core/Utilities/ArrayUtils.cs:79

                    limCur = mid;
                else
                    minCur = mid + 1;

                Debug.Assert(min <= minCur && minCur <= limCur && limCur <= lim);
                Debug.Assert(minCur == min || input[minCur - 1] < value);
                Debug.Assert(limCur == lim || input[limCur] >= value);
            }
            Debug.Assert(min <= minCur && minCur == limCur && limCur <= lim);
            Debug.Assert(minCur == min || input[minCur - 1] < value);
            Debug.Assert(limCur == lim || input[limCur] >= value);

            return minCur;
        }

        public static int EnsureSize<T>(ref T[] array, int min, int max, bool keepOld, out bool resized)
        {
            if (min > max)
                throw new ArgumentOutOfRangeException(nameof(max), "min must not exceed max");

            // This code adapted from the private method EnsureCapacity code of List<T>.
            int size = ArrayUtils.Size(array);
            if (size >= min)
            {
                resized = false;
                return size;
            }

            int newSize = size == 0 ? 4 : size * 2;
            // This constant taken from the internal code of system\array.cs of mscorlib.
            if ((uint)newSize > max)
                newSize = max;
            if (newSize < min)
                newSize = min;
            if (keepOld && size > 0)
                Array.Resize(ref array, newSize);
            else

View on GitHub (pinned to 7b76e69cf9)

Solutions

Verify the min/max computation: ensure max is always >= min before the call, e.g. use Math.Max(min, currentSize) for the max parameter.
If you intended simple growth, pass the same value or a larger value for max (or the maximum allowed size) rather than a stale capacity value.
Guard the call: if (min > max) swap or clamp the values before invoking EnsureSize.

Example fix

// before
ArrayUtils.EnsureSize(ref buffer, requiredSize, currentCapacity, true, out resized);
// after
int max = Math.Max(requiredSize, currentCapacity);
ArrayUtils.EnsureSize(ref buffer, requiredSize, max, true, out resized);

Defensive patterns

Strategy: validation

Validate before calling

if (min > max)
    max = Math.Max(min, max); // or throw your own descriptive error
ArrayUtils.EnsureSize(ref array, min, max, keepOld, out resized);

Try / catch

try
{
    ArrayUtils.EnsureSize(ref array, min, max, keepOld, out resized);
}
catch (ArgumentOutOfRangeException)
{
    // min > max: recompute bounds from a single source of truth
}

Prevention

Derive min and max from one shared computation to keep them consistent.
Add a debug assert (Debug.Assert(min <= max)) at call sites.
Watch for integer overflow when computing large sizes.

When it happens

Trigger: Calling ArrayUtils.EnsureSize(ref arr, min, max, keepOld, out resized) with min > max, e.g. computing min and max from two different length calculations that disagree, or passing a fixed max (like a current capacity) with a min computed from a larger required size.

Common situations: Custom data loaders or transforms in ML.NET pipelines that pre-compute required buffer sizes; off-by-one or integer-overflow bugs when computing new sizes; passing ArrayUtils.Size(oldArray) as max while min comes from a growing dataset.

Related errors

AI-assisted analysis of dotnet/machinelearning@7b76e69cf9 (2026-09-11). Data as JSON: /api/errors/52e2be46452dd559. Report an issue: GitHub.