prestodb/presto · error · PrestoException

NOT_SUPPORTED

NOT_SUPPORTED

Error message

Unsupported coercion from %s to %s

What it means

HiveCoercionRecordCursor.createCoercer is the cursor-level twin of HiveCoercer.createCoercer, used when reading old files after a column type change and bridging values through a RecordCursor; it supports integer upscaling, varchar<->integer, float->double, and recursive list/map/struct coercions. Any other from/to combination throws NOT_SUPPORTED, since Presto cannot convert the on-disk values to the new declared column type.

Source

Thrown at presto-hive/src/main/java/com/facebook/presto/hive/HiveCoercionRecordCursor.java:313

            return new IntegerNumberUpscaleCoercer();
        }
        else if (fromHiveType.equals(HIVE_INT) && toHiveType.equals(HIVE_LONG)) {
            return new IntegerNumberUpscaleCoercer();
        }
        else if (fromHiveType.equals(HIVE_FLOAT) && toHiveType.equals(HIVE_DOUBLE)) {
            return new FloatToDoubleCoercer();
        }
        else if (isArrayType(fromType) && isArrayType(toType)) {
            return new ListCoercer(typeManager, fromHiveType, toHiveType, bridgingRecordCursor);
        }
        else if (isMapType(fromType) && isMapType(toType)) {
            return new MapCoercer(typeManager, fromHiveType, toHiveType, bridgingRecordCursor);
        }
        else if (isRowType(fromType) && isRowType(toType)) {
            return new StructCoercer(typeManager, fromHiveType, toHiveType, bridgingRecordCursor);
        }

        throw new PrestoException(NOT_SUPPORTED, format("Unsupported coercion from %s to %s", fromHiveType, toHiveType));
    }

    private static class IntegerNumberUpscaleCoercer
            extends Coercer
    {
        @Override
        public void coerce(RecordCursor delegate, int field)
        {
            setLong(delegate.getLong(field));
        }
    }

    private static class IntegerNumberToVarcharCoercer
            extends Coercer
    {
        @Override
        public void coerce(RecordCursor delegate, int field)
        {

View on GitHub (pinned to 55bb57d202)

Solutions

  1. Physically rewrite the data with explicit casts: CREATE TABLE t_new AS SELECT CAST(col AS <new_type>) ... then swap, so no read-time coercion is needed.
  2. Restrict the ALTER to coercion-supported transitions (tinyint/smallint/int->larger ints, int types->varchar, varchar->integer types, float->double, and element-wise compatible array/map/struct changes).
  3. For nested types, evolve each element/field type only to supported coercions (e.g. array<int> -> array<bigint>), otherwise rewrite the column via INSERT OVERWRITE with casts.

Example fix

-- before: unsupported read-time coercion after ALTER
ALTER TABLE t CHANGE COLUMN s s DECIMAL(10,2);
SELECT * FROM t; -- fails: Unsupported coercion from varchar to decimal(10,2)

-- after: rewrite data with explicit cast
CREATE TABLE t_new AS SELECT CAST(s AS DECIMAL(10,2)) AS s FROM t;
-- drop t, rename t_new to t
Defensive patterns

Strategy: validation

Validate before calling

-- before changing a column type that old files must read through, verify the pair is supported:
-- integer upscale (tinyint/smallint/int -> larger ints), int<->varchar, float -> double,
-- array->array / map->map / row->row with supported element/field coercions.
-- Otherwise plan a CTAS rewrite with explicit CASTs.

Try / catch

try { ResultSet rs = stmt.executeQuery("SELECT * FROM evolved_table"); ... }
catch (SQLException e) {
  if (e.getMessage() != null && e.getMessage().contains("Unsupported coercion from")) {
    // rewrite old partitions with explicit casts, then retry the query
  } else throw e;
}

Prevention

When it happens

Trigger: Querying a Hive table after ALTER TABLE changed a column to a type outside the supported coercion set (e.g. string->date/timestamp/boolean/decimal, int->double/real, tinyint->float), where the read path goes through HiveCoercionRecordCursor with a bridging record cursor; also unsupported nested element/field coercions inside array/map/struct columns.

Common situations: Schema evolution altering string columns to date/decimal/boolean; converting int to double/float; struct field type changes to unsupported targets; then running a SELECT over partitions still containing files written with the old schema.

Understand the failure class

Background: Presto NOT_SUPPORTED error: what "not supported" means and how to fix it — this error's family across 3 libraries.

Related errors

AI-assisted analysis of prestodb/presto@55bb57d202 (2026-09-04). Data as JSON: /api/errors/087d316cab924efa. Report an issue: GitHub.