Skip to main content

Int

function Int(E: Extended): Extended;

Example​

procedure ScriptEvent(var Value: variant);
begin
  Value := Int(-12.875); // -12.0
end;

Usage​

Returns the integer part of an Extended value by rounding toward zero while retaining a floating-point result.

Parameters

NameTypeDescription
EExtendedValue whose fractional component is to be discarded.

Returns

The integer-valued Extended nearest to E in the direction of zero.

InputResult
12.87512.0
-12.875-12.0
0.750.0
-0.75zero, with signed-zero representation dependent on the RTL path

This is truncation, not floor: Int(-12.875) is -12, not -13.

Additional Technical Info​

Int removes the fractional part of a real value by rounding toward zero. Unlike Trunc, it returns an Extended floating-point result rather than an integer result.

This is the PascalScript standard function and delegates directly to Delphi's System.Int. The example is fictional and source-reviewed only.

Implementation

  1. The modified PascalScript compiler declares Int in its standard Math group.
  2. Runtime selector 20 reads the argument through Stack.GetReal.
  3. Delphi System.Int rounds toward zero while retaining a floating-point return type.
  4. The result is written back with Stack.SetReal.

The product project targets Win32 and Win64 and currently defaults to Win64. The Win32 RTL temporarily selects x87 round-toward-zero mode for this operation and restores the original control word. The Win64 path uses its SSE/double-precision implementation.

Edge cases and quirks

  • A value already having no representable fractional part is returned unchanged.
  • Sufficiently large floating-point values have no stored fractional bits even when their source decimal text appeared to contain a fraction.
  • The result remains floating point. Use it when subsequent calculations need Extended; use Trunc when an integer return and its range rules are required.
  • Infinity and NaN are floating-point special values rather than ordinary integers. Their propagation or exception behaviour depends on the RTL target and active exception mask.
  • Signed zero is numerically equal to zero. Do not use this function when preservation of a negative-zero sign bit is a business requirement.

Errors and side effects

The operation is pure. The wrapper performs no range check or exception translation. Exceptional floating-point input can signal through Delphi according to the process floating-point environment.

Related entries

  • Abs removes a sign without removing a fraction.
  • Trunc also rounds toward zero but returns an integer.

External references

Created 2026-07-15