pathwaycom/pathway · error · TypeError

Pathway does not support using reducer {self} on column of t

Error message

Pathway does not support using reducer {self} on column of type {arg_type}.

What it means

The sum reducer (pathway.reducers.sum) only accepts columns whose dtype is a float subtype or an array subtype (checked via dt.dtype_issubclass against dt.FLOAT and ANY_ARRAY; ints are handled separately in engine_reducer_unary). If the aggregated column is e.g. str, bool, Pointer, or a duration/complex type, return_type_unary raises this TypeError at expression-building time.

Source

Thrown at python/pathway/internals/reducers.py:120

        warnings.warn(
            f"{self.name} reducer uses processing time to choose elements"
            + " while windowby uses data time to assign entries to windows."
            + " Maybe it is not the behavior you want. To choose elements according"
            + f" to their data time, you may use {self.alternative} reducer.",
            stacklevel=12,
        )


class SumReducer(UnaryReducer):
    def __init__(self, name: str, strict: bool) -> None:
        super().__init__(name=name)
        self.strict = strict

    def return_type_unary(self, arg_type: dt.DType, id_type: dt.DType) -> dt.DType:
        for allowed_dtype in [dt.FLOAT, dt.ANY_ARRAY]:
            if dt.dtype_issubclass(arg_type, allowed_dtype):
                return arg_type
        raise TypeError(
            f"Pathway does not support using reducer {self}"
            + f" on column of type {arg_type}.\n"
        )

    def engine_reducer_unary(self, arg_type: dt.DType) -> api.Reducer:
        if arg_type == dt.INT:
            if self.strict:
                raise ValueError(
                    "Setting strict=True in pathway.reducers.sum when the column has type int is not allowed"
                )
            return api.Reducer.INT_SUM
        elif isinstance(arg_type, dt.Array):
            return api.Reducer.array_sum(self.strict)
        else:
            return api.Reducer.float_sum(self.strict)


class SortedTupleWrappingReducer(UnaryReducerWithDefault):

View on GitHub (pinned to fa2f74a464)

Solutions

1. Cast the column to numeric before reducing: pw.reducers.sum(pw.this.col.astype(float)) (or int, which selects INT_SUM).
2. Fix ingestion so the column is parsed as int/float (e.g. input_format / schema with dtype=float in pw.io.csv.read or pw.io.jsonl.read).
3. If the column is genuinely non-numeric, use a different reducer (tuple, count, min/max on comparable types) — summing text is not supported.

Example fix

# before
agg = table.groupby(pw.this.key).reduce(total=pw.reducers.sum(pw.this.price))  # price is str

# after
agg = table.groupby(pw.this.key).reduce(total=pw.reducers.sum(pw.this.price.astype(float)))

import pathway as pw
from pathway.internals import dtype as dt

def column_is_summable(table: pw.Table, name: str) -> bool:
    d = table.schema._dtypes()[name]
    return dt.dtype_issubclass(d, dt.FLOAT) or dt.dtype_issubclass(d, dt.ANY_ARRAY) or d == dt.INT

import pathway as pw
from pathway.internals import dtype as dt

def is_numeric_column(table: pw.Table, name: str) -> bool:
    d = table.schema._dtypes()[name]
    return dt.dtype_issubclass(d, dt.FLOAT) or d == dt.INT

When it happens

Trigger: table.reduce(s=pw.reducers.sum(t.col)) where col has dtype str, bool, Optional[str], json, or any non-numeric non-array type; equivalently table.groupby(...).reduce(sum=pw.reducers.sum(...)) on a text column.

Common situations: CSV columns auto-typed as str that hold numbers ('price' parsed as string); aggregating a json column; trying to sum boolean flags.

Related errors

• Pathway does not support using reducer {self.name} on column
• Pathway does not support using binary operator {expression._
• Incompatible types in for a binary operator. The types are:
• {role} {col._name!r} must be of type str, got {col._column.d
• Cannot flatten column of type {dtype}.

AI-assisted analysis of pathwaycom/pathway@fa2f74a464 (2026-08-15). Data as JSON: /api/errors/4b203b59d2679823. Report an issue: GitHub.