pathwaycom/pathway · error · ValueError

Only one column from a table can be used in a…

Error message

Only one column from a table can be used in a synchronization group

What it means

ValueError raised when the synchronization code finds an InputOperator for the column's table whose data source already has a synchronization_group assigned. Each input table (connector) can contribute at most one column to synchronization groups, because the group is attached to the data source options, not per column.

Solutions

Use only one column per table across all synchronization groups; pick the primary timestamp (e.g. unix_timestamp)
If a second key of the same source must be synchronized, feed that source through a second independent connector/table
Audit all sync() calls and ensure no table contributes more than one column reference

Example fix

// before
pw.io.synchronize(t.created_at, u.time, max_difference=60)
pw.io.synchronize(t.updated_at, v.time, max_difference=60)  # ValueError: same table twice

// after
pw.io.synchronize(t.created_at, u.time, v.time, max_difference=60)

Defensive patterns

Strategy: validation

Validate before calling

from collections import Counter

def validate_one_column_per_table(columns) -> None:
    tables = Counter(c._table for c in columns)
    dupes = [t for t, n in tables.items() if n > 1]
    if dupes:
        raise ValueError(f'tables used with more than one sync column: {dupes}')

Prevention

Track which tables already joined a sync group in pipeline builders
Design one timestamp key per connector for synchronization
Add a lint/test pass asserting each input table appears in at most one group

When it happens

Trigger: Calling sync() twice with different columns of the same table: sync(t.a, u.x, ...) then sync(t.b, v.y, ...); or passing two columns of the same table in one call. The second assignment hits data_source_options.synchronization_group is not None.

Common situations: Trying to synchronize two connectors by two different time keys of the same source (e.g. created_at and updated_at); composing multiple sync groups in a larger pipeline where one table appears twice; refactoring that accidentally duplicates a table's column into another group.

Related errors

AI-assisted analysis of pathwaycom/pathway@fa2f74a464 (2026-08-15). Data as JSON: /api/errors/69e0510c0579883f. Report an issue: GitHub.

Appendix: source

Thrown at python/pathway/io/_synchronization.py:291

        if column_idx is None:
            raise ValueError(
                f"Failed to find the column '{column._name}' in table {column._table}"
            )

        is_table_found = False
        for node in G._current_scope.nodes:
            if (
                not isinstance(node, InputOperator)
                or not isinstance(node.datasource, GenericDataSource)
                or node.outputs[0].value != column._table
            ):
                continue
            is_table_found = True
            group = api.ConnectorGroupDescriptor(
                name, column_idx, max_difference, priority, idle_duration
            )
            if node.datasource.data_source_options.synchronization_group is not None:
                raise ValueError(
                    "Only one column from a table can be used in a synchronization group"
                )
            node.datasource.data_source_options.set_synchronization_group(group)
            break

        if not is_table_found:
            raise ValueError(
                "Only unchanged columns of an input tables can be used in input synchronization groups"
            )

    if len(column_types) > 1:
        raise ValueError(
            "All synchronization group column types must coincide. "
            "However several types have been detected: {}".format(
                ", ".join(sorted([f"'{t}'" for t in column_types]))
            )
        )

View on GitHub (pinned to fa2f74a464)