[Feature] Support Spark expression: make_dt_interval #3098

What is the problem the feature request solves?

Note: This issue was generated with AI assistance. The specification details have been extracted from Spark documentation and may need verification.

Comet does not currently support the Spark make_dt_interval function, causing queries using this function to fall back to Spark's JVM execution instead of running natively on DataFusion.

The MakeDTInterval expression creates a day-time interval value from separate day, hour, minute, and second components. This expression is used to construct DayTimeIntervalType values programmatically by combining individual time unit values into a single interval representation.

Supporting this expression would allow more Spark workloads to benefit from Comet's native acceleration.

Describe the potential solution

Spark Specification

Syntax:

make_dt_interval(days, hours, minutes, seconds)
make_dt_interval(days, hours, minutes)
make_dt_interval(days, hours)
make_dt_interval(days)
make_dt_interval()

Arguments:

Return Type: Returns a DayTimeIntervalType() representing the constructed interval.

Supported Data Types:

• days: Integer values
• hours: Integer values
• minutes: Integer values
• seconds: Decimal values with up to 6 decimal places for microsecond precision

Edge Cases:

• Null handling: Expression is null-intolerant (nullIntolerant = true), meaning if any input is null, the result is null
• Default values: Missing parameters default to 0 (literal values)
• Precision handling: Seconds parameter uses DecimalType with 6 decimal places to preserve microsecond precision
• Overflow behavior: Delegates to IntervalUtils.makeDayTimeInterval() for overflow validation and error handling
• Error context: Includes query context information for meaningful error messages when interval construction fails

Examples:

-- Create a 5-day, 3-hour, 30-minute, 45.5-second interval
SELECT make_dt_interval(5, 3, 30, 45.5);

-- Create a 2-day interval
SELECT make_dt_interval(2);

-- Create a 1-day, 12-hour interval  
SELECT make_dt_interval(1, 12);

-- Create an empty interval
SELECT make_dt_interval();

// DataFrame API usage
import org.apache.spark.sql.functions._

// Create interval from literal values
df.select(expr("make_dt_interval(5, 3, 30, 45.5)"))

// Create interval from column values
df.select(expr("make_dt_interval(day_col, hour_col, min_col, sec_col)"))

Implementation Approach

See the Comet guide on adding new expressions for detailed instructions.

1. Scala Serde: Add expression handler in spark/src/main/scala/org/apache/comet/serde/
2. Register: Add to appropriate map in QueryPlanSerde.scala
3. Protobuf: Add message type in native/proto/src/proto/expr.proto if needed
4. Rust: Implement in native/spark-expr/src/ (check if DataFusion has built-in support first)

Additional context

Difficulty: Large
Spark Expression Class: org.apache.spark.sql.catalyst.expressions.MakeDTInterval

Related:

• MakeYMInterval - Creates year-month intervals
• IntervalUtils - Utility class for interval operations
• DayTimeIntervalType - The data type returned by this expression
• Extract - Extracts components from interval values

This issue was auto-generated from Spark reference documentation.