handle_overflow

datafusion_expr_common::interval_arithmetic

Function handle_overflow 

Source 
fn handle_overflow<const UPPER: bool>(
    dt: &DataType,
    op: Operator,
    lhs: &ScalarValue,
    rhs: &ScalarValue,
) -> ScalarValue
Expand description

This function handles cases where an operation results in an overflow. Such results are converted to an unbounded endpoint if:

  • We are calculating an upper bound and we have a positive overflow.
  • We are calculating a lower bound and we have a negative overflow.

Otherwise, the function sets the endpoint as:

  • The minimum representable number with the given datatype (dt) if we are calculating an upper bound and we have a negative overflow.
  • The maximum representable number with the given datatype (dt) if we are calculating a lower bound and we have a positive overflow.

Caution: This function contains multiple calls to unwrap(), and may return non-standardized interval bounds. Therefore, it should be used with caution. Currently, it is used in contexts where the DataType (dt) is validated prior to calling this function, op is supported by interval library, and the following interval creation is standardized with Interval::new.