pathwaycom/pathway · error · TypeError

Schemas should not be called. Use `table.schema` not `table.

Error message

Schemas should not be called. Use `table.schema` not `table.schema()`.

What it means

pw.Schema metaclass defines __call__ to always raise, because a Schema is a class (type definition), not an instance you construct or call. This guard catches the common mistake of treating table.schema like a method — the schema is a plain attribute in Pathway, unlike older APIs or pandas-style accessors.

Source

Thrown at python/pathway/internals/schema.py:318

        self.__columns__ = _create_column_definitions(self, bases, schema_properties)
        pk_dtypes = [col.dtype for col in self.__columns__.values() if col.primary_key]
        if len(pk_dtypes) > 0:
            derived_type = dt.Pointer(*pk_dtypes)
            assert id_dtype in [derived_type, dt.ANY_POINTER]
            id_dtype = derived_type
        self.__universe_properties__ = _universe_properties(
            list(self.__columns__.values()),
            schema_properties,
            dtype=id_dtype,
            append_only=id_append_only,
        )
        self.__dtypes__ = {
            name: column.dtype for name, column in self.__columns__.items()
        }
        self.__types__ = {k: v.typehint for k, v in self.__dtypes__.items()}

    def __call__(self) -> NoReturn:
        raise TypeError(
            "Schemas should not be called. Use `table.schema` not `table.schema()."
        )

    def __or__(self, other: type[Schema]) -> type[Schema]:  # type: ignore
        return schema_add(self, other)  # type: ignore

    def columns(self) -> Mapping[str, ColumnSchema]:
        return MappingProxyType(self.__columns__)

    def column_names(self) -> list[str]:
        return list(self.keys())

    def columns_to_json_serializable_dict(self) -> dict:
        columns = {}
        for column_name, column_schema in self.columns().items():
            columns[column_name] = column_schema.to_json_serializable_dict()
        return columns

View on GitHub (pinned to fa2f74a464)

Solutions

1. Use table.schema without parentheses: table.schema.columns(), table.schema.typehints(), etc.
2. To create a typed empty table use pw.Table.empty(schema=MySchema) (or pw.debug.table_from_markdown for tests), not MySchema().

Example fix

# before
cols = table.schema().columns()

# after
cols = table.schema.columns()

import pathway as pw

def schema_usage_ok(table: pw.Table) -> None:
    assert not callable(table.schema), "use table.schema (no parentheses)"

When it happens

Trigger: Calling table.schema() (with parentheses) instead of using table.schema; also attempting to 'instantiate' a schema class like MySchema() to build an empty table.

Common situations: Migrating from Pathway versions or tutorials where schema() was callable; muscle memory from accessor methods; using table.schema().columns instead of table.schema.columns.

Related errors

• `{self.name}` has no attribute `{name}`
• Schema.with_types() argument name has to be an existing colu
• Schema.without() argument {name!r} has to refer to an existi
• Failed to detect the region of S3 bucket {bucket!r} (HTTP st
• SchemaRegistryHeader.value must be a str, got {type(self.val

AI-assisted analysis of pathwaycom/pathway@fa2f74a464 (2026-08-15). Data as JSON: /api/errors/079042d89b53236f. Report an issue: GitHub.