Skip to main content

SecondSpan

Function SecondSpan( const ANow, AThen : TDateTime) : Double

Example​

procedure ScriptEvent(var Value: variant);
var
  EarlierValue, LaterValue: TDateTime;
begin
  EarlierValue := EncodeDateTime(2026, 7, 16, 9, 0, 0, 0);
  LaterValue := EncodeDateTime(2026, 7, 16, 9, 1, 30, 500);
  Value := SecondSpan(LaterValue, EarlierValue); // approximately 90.5
end;

Usage​

SecondSpan returns the absolute fractional-second span from Velox's raw numeric TDateTime difference.

Parameters

NameTypeDescription
ANowTDateTime, constOne endpoint; it need not be later.
AThenTDateTime, constOther endpoint.

Returns

A non-negative Double containing whole and fractional seconds.

Behaviour

  • Argument order is discarded through the absolute raw difference.
  • Fractional seconds are retained subject to floating-point precision.
  • No current-clock, timezone or daylight-saving operation occurs.
  • For ordinary post-epoch values, the result represents their numeric day difference in seconds.

Errors

No validation is performed. Non-finite or extreme raw values can produce undefined/exceptional floating behaviour.

Additional Technical Info​

SecondSpan returns an absolute fractional number of seconds by subtracting the underlying floating TDateTime numbers and multiplying the raw day span by 86,400.

The example is fictional, deterministic and source-reviewed. It was not executed by the documentation workflow.

Implementation

The DateUtils import binds to System.DateUtils.SecondSpan. The installed routine calls SpanOfNowAndThen, which orders the raw floating values and subtracts the smaller from the larger, then multiplies by SecsPerDay (86,400). It does not convert either endpoint to a timestamp.

Edge cases and quirks

  • Delphi encodes the time portion of a pre-epoch TDateTime with a negative fractional sign. Raw subtraction therefore does not preserve calendar chronology for timed pre-epoch pairs or many cross-epoch pairs. SecondSpan can return the wrong span for those valid values.
  • Use SecondsBetween for timestamp-normalised complete seconds. It avoids this pre-epoch defect but truncates the fractional remainder.
  • Binary floating-point calculation can produce a result slightly above or below an expected decimal; use an appropriate tolerance.
  • Free Pascal's same-name API is useful compatibility context but does not override the installed Delphi implementation.

Side effects

None.

Performance and concurrency

Constant-time floating comparison, subtraction and multiplication with no shared state.

Remarks

This routine is suitable when fractional precision is more important than complete-unit truncation and all values are in a safe post-epoch range.

Related entries

External references

Created 2026-07-15