apache/beam · error · UnsupportedOperationException

[ ] is not supported in AVG

Error message

[%s] is not supported in AVG

What it means

createAvg builds the CombineFn for SQL AVG, supporting INTEGER-family (FloatAvg/IntAvg), DOUBLE (DoubleAvg), and DECIMAL (BigDecimalAvg). Any other Schema.FieldType hits the default branch and throws UnsupportedOperationException naming the type. AVG requires a numeric type with a supported mean combiner.

Solutions

  1. Cast the column to DOUBLE or DECIMAL in the query: SELECT AVG(CAST(ts AS DOUBLE)) ...
  2. Restrict AVG to numeric columns; compute averages for other types manually with a Map/Combine pipeline.
  3. Supply a custom CombineFn computing sum/count for the desired type.

Example fix

// before
CombineFn fn = BeamBuiltinAggregations.createAvg(Schema.FieldType.STRING); // throws

// after
CombineFn fn = BeamBuiltinAggregations.createAvg(Schema.FieldType.DOUBLE);
// SQL: SELECT AVG(CAST(score AS DOUBLE)) FROM t
Defensive patterns

Strategy: validation

Validate before calling

java.util.Set<Schema.TypeName> ok =
    java.util.Set.of(Schema.TypeName.INTEGER, Schema.TypeName.INT64, Schema.TypeName.DOUBLE, Schema.TypeName.DECIMAL);
if (!ok.contains(fieldType.getTypeName())) {
  throw new IllegalArgumentException("AVG unsupported for " + fieldType);
}

Type guard

boolean avgSupported(Schema.FieldType t) {
  switch (t.getTypeName()) {
    case INTEGER: case INT64: case DOUBLE: case DECIMAL: return true;
    default: return false;
  }
}

Try / catch

try {
  CombineFn fn = BeamBuiltinAggregations.createAvg(fieldType);
} catch (UnsupportedOperationException e) {
  // fall back to CAST(col AS DOUBLE) then AVG, or custom sum/count combiner
}

Prevention

When it happens

Trigger: Calling BeamBuiltinAggregations.createAvg(fieldType) with a non-numeric or unsupported numeric TypeName (STRING, BOOLEAN, DATE, etc.) — typically from SELECT AVG(varchar_col) or programmatic pipeline construction with a wrong field type.

Common situations: AVG over a string or boolean column from a query mistake; AVG over TIMESTAMP/DATE which many engines allow but this implementation does not; schema inferred as DECIMAL variant not covered by the switch.

Related errors

AI-assisted analysis of apache/beam@12126d8942 (2026-09-13). Data as JSON: /api/errors/a81520423c88c382. Report an issue: GitHub.

Appendix: source

Thrown at sdks/java/extensions/sql/src/main/java/org/apache/beam/sdk/extensions/sql/impl/transform/BeamBuiltinAggregations.java:227

  /** {@link CombineFn} for AVG. */
  static CombineFn createAvg(Schema.FieldType fieldType) {
    switch (fieldType.getTypeName()) {
      case INT32:
        return new IntegerAvg();
      case INT16:
        return new ShortAvg();
      case BYTE:
        return new ByteAvg();
      case INT64:
        return new LongAvg();
      case FLOAT:
        return new FloatAvg();
      case DOUBLE:
        return new DoubleAvg();
      case DECIMAL:
        return new BigDecimalAvg();
      default:
        throw new UnsupportedOperationException(
            String.format("[%s] is not supported in AVG", fieldType));
    }
  }

  static CombineFn createBitOr(Schema.FieldType fieldType) {
    if (fieldType.getTypeName() == TypeName.INT64) {
      return new BitOr();
    }
    throw new UnsupportedOperationException(
        String.format("[%s] is not supported in BIT_OR", fieldType));
  }

  static CombineFn createBitAnd(Schema.FieldType fieldType) {
    if (fieldType.getTypeName() == TypeName.INT64) {
      return new BitAnd();
    }
    throw new UnsupportedOperationException(
        String.format("[%s] is not supported in BIT_AND", fieldType));

View on GitHub (pinned to 12126d8942)