dotnet/machinelearning · error · ArgumentException

String.Format(Strings.MultipleMismatchedValueType…

Error message

String.Format(Strings.MultipleMismatchedValueType, typeof(long), typeof(int), typeof(bool))

What it means

When Clone receives a non-null mapIndices column, its DataType must be long, int, or bool; any other element type throws ArgumentException with Strings.MultipleMismatchedValueType listing the three supported types, paramName mapIndices. Only these three map column types have cloning implementations.

Solutions

1. Convert the map to PrimitiveDataFrameColumn<long> (or int/bool) before cloning.
2. Build the map explicitly: new PrimitiveDataFrameColumn<long>("indices", indices).
3. Keep filter maps bool and index maps long by convention.
4. Check mapIndices.DataType against typeof(long)/typeof(int)/typeof(bool) before calling Clone.

Example fix

// before
column.Clone(floatIndices); // PrimitiveDataFrameColumn<float>
// after
var longIndices = new PrimitiveDataFrameColumn<long>("indices",
    Enumerable.Range(0, (int)floatIndices.Length).Select(i => (long)floatIndices[i]));
column.Clone(longIndices);

var t = mapIndices?.DataType;
bool ok = t == typeof(long) || t == typeof(int) || t == typeof(bool);
if (mapIndices != null && !ok) throw new ArgumentException("mapIndices must be long, int, or bool");

bool isValidMapIndices(DataFrameColumn m) =>
    m is null || m.DataType == typeof(long) || m.DataType == typeof(int) || m.DataType == typeof(bool);

try { var clone = column.Clone(mapIndices); }
catch (ArgumentException ex) when (ex.ParamName == "mapIndices")
{ /* convert mapIndices to PrimitiveDataFrameColumn<long> and retry */ }

When it happens

Trigger: Calling Clone(mapIndices, invertMapIndices) with a PrimitiveDataFrameColumn of another type (short, byte, float, decimal, string) or any non-primitive column.

Common situations: Index lists produced elsewhere as float/double; CSV-parsed indices typed wrong; generic helper code passing whatever column it holds.

Understand the failure class

Background: UnsupportedOperationException and "is not supported" errors: when a library deliberately refuses a call — this error's family across 30 libraries.

Related errors

• Strings.MapIndicesExceedsColumnLength
• and must be different
• nameof(groupByColumnIndex)
• string.Format(Strings.DuplicateColumnName, column.Name)
• string.Format(Strings.MismatchedValueType, typeof

AI-assisted analysis of dotnet/machinelearning@7b76e69cf9 (2026-09-11). Data as JSON: /api/errors/954809b473d6fe73. Report an issue: GitHub.

Appendix: source

Thrown at src/Microsoft.Data.Analysis/DataFrameColumns/VBufferDataFrameColumn.cs:287

        public new VBufferDataFrameColumn<T> Clone(DataFrameColumn mapIndices, bool invertMapIndices, long numberOfNullsToAppend)
        {
            return (VBufferDataFrameColumn<T>)CloneImplementation(mapIndices, invertMapIndices, numberOfNullsToAppend);
        }

        public new VBufferDataFrameColumn<T> Clone(long numberOfNullsToAppend = 0)
        {
            return (VBufferDataFrameColumn<T>)CloneImplementation(numberOfNullsToAppend);
        }

        protected override DataFrameColumn CloneImplementation(DataFrameColumn mapIndices, bool invertMapIndices = false, long numberOfNullsToAppend = 0)
        {
            VBufferDataFrameColumn<T> clone;
            if (!(mapIndices is null))
            {
                Type dataType = mapIndices.DataType;
                if (dataType != typeof(long) && dataType != typeof(int) && dataType != typeof(bool))
                    throw new ArgumentException(String.Format(Strings.MultipleMismatchedValueType, typeof(long), typeof(int), typeof(bool)), nameof(mapIndices));
                if (mapIndices.DataType == typeof(long))
                    clone = CloneImplementation(mapIndices as PrimitiveDataFrameColumn<long>, invertMapIndices);
                else if (dataType == typeof(int))
                    clone = CloneImplementation(mapIndices as PrimitiveDataFrameColumn<int>, invertMapIndices);
                else
                    clone = CloneImplementation(mapIndices as PrimitiveDataFrameColumn<bool>);
            }
            else
            {
                clone = Clone();
            }

            return clone;
        }

        protected override DataFrameColumn CloneImplementation(long numberOfNullsToAppend)
        {
            var ret = new VBufferDataFrameColumn<T>(Name, Length);

View on GitHub (pinned to 7b76e69cf9)