apache/iceberg · error · UnsupportedOperationException

Not an long column

Error message

Not an long column

What it means

TripleWriter.writeLong's default body throws UnsupportedOperationException with a (typo'd) message 'Not an long column'. Only writers for INT64 Parquet columns override it; all other column writers keep the throwing default, so calling writeLong on a non-INT64 writer fails at runtime.

Source

Thrown at parquet/src/main/java/org/apache/iceberg/parquet/TripleWriter.java:61

  /**
   * Write a triple.
   *
   * @param rl repetition level
   * @param value the integer value
   */
  default void writeInteger(int rl, int value) {
    throw new UnsupportedOperationException("Not an integer column");
  }

  /**
   * Write a triple.
   *
   * @param rl repetition level
   * @param value the long value
   */
  default void writeLong(int rl, long value) {
    throw new UnsupportedOperationException("Not an long column");
  }

  /**
   * Write a triple.
   *
   * @param rl repetition level
   * @param value the float value
   */
  default void writeFloat(int rl, float value) {
    throw new UnsupportedOperationException("Not an float column");
  }

  /**
   * Write a triple.
   *
   * @param rl repetition level
   * @param value the double value
   */

View on GitHub (pinned to 86d9c8fc54)

Solutions

1. Verify the Parquet column physical type is INT64 before calling writeLong.
2. Override writeLong in custom TripleWriter implementations for INT64 columns.
3. Check the type mapping produced by TypeToMessageType for the field (timestamps must map to INT64 with a logical annotation).

Example fix

// before
writer.writeLong(rl, value); // column is INT32
// after
writer.writeInteger(rl, (int) value); // or fix the schema so the column is INT64

boolean isInt64Column(ColumnDescriptor col) {
  return col.getPrimitiveType().getPrimitiveTypeName() == PrimitiveTypeName.INT64;
}

if (col.getPrimitiveType().getPrimitiveTypeName() == PrimitiveTypeName.INT64) { writer.writeLong(rl, value); }

When it happens

Trigger: Calling column.writeLong(rl, value) on a TripleWriter whose Parquet column is not INT64 (e.g. INT32, DOUBLE, BINARY).

Common situations: Writing Iceberg long or timestamp values through a writer for an int column; custom TripleWriter subclasses missing the writeLong override; schema evolution (int->long) not mirrored in writers.

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

• Unsupported type: %s
• Unsupported type:
• Not a boolean column
• Not an integer column
• Not an float column

AI-assisted analysis of apache/iceberg@86d9c8fc54 (2026-09-12). Data as JSON: /api/errors/abd8d6cc37d670e1. Report an issue: GitHub.