Skip to main content

Abs

function Abs(E: Extended): Extended;

Example​

procedure ScriptEvent(var Value: variant);
begin
  Value := Abs(-125.75); // 125.75
end;

Usage​

Abs returns the non-negative magnitude of an Extended floating-point value by clearing its sign.

Parameters

NameTypeDescription
EExtendedFloating-point value whose sign is to be removed. Integer arguments are converted to Extended for this declaration.

Returns

The same floating-point magnitude as E, with the sign cleared. The return type remains Extended; this entry does not return an integer and does not round the value.

Additional Technical Info​

Abs returns the magnitude of an Extended value. Negative finite input becomes positive, while positive input and positive zero are unchanged.

This is the PascalScript standard function, not a Velox-specific numeric helper. The example is fictional and source-reviewed only.

Implementation

  1. The modified PascalScript compiler declares Abs in its standard Math group.
  2. At runtime, standard-procedure selector 22 reads the argument through Stack.GetReal.
  3. It calls Delphi's intrinsic System.Abs and writes the resulting real value to the PascalScript stack.

There is no Velox validation, conversion to Variant, configuration access or flow-state change.

Edge cases and quirks

  • Abs(-0.0) is positive zero because the Delphi floating-point implementation clears the sign bit.
  • Positive and negative infinity both produce positive infinity.
  • A NaN remains a NaN. Clearing the sign can change its sign bit, but does not make it a number or make equality comparisons useful.
  • The overload visible to Velox accepts Extended. It is different from Delphi's integer overloads, so integer minimum-value overflow associated with fixed-width integer absolute value does not apply here.
  • Extended precision follows the host Delphi target. In targets where Extended aliases Double, retaining the Extended name does not add 80-bit precision.

Errors and side effects

The operation is pure and normally does not raise for finite values, infinities or a quiet NaN. A signalling NaN or a non-default floating-point environment can still produce a hardware floating-point exception. Velox does not catch or translate such an exception in this wrapper.

Related entries

  • Sqrt returns a principal square root rather than merely removing a sign.
  • Int removes a fractional part by rounding toward zero.

External references

Created 2026-07-15