Skip to main content

vxFloatToStr

function vxFloatToStr(Value: Extended): string;

Example​

procedure ScriptEvent(var Value: variant);
var
  Reading: Extended;
begin
  Reading := 125.5;
  Value := vxFloatToStr(Reading);
end;

Usage​

vxFloatToStr formats an Extended value as text with Windows system-locale settings.

Parameters

NameTypeDescription
ValueExtendedFloating-point value to format. Actual precision and range depend on the Velox target.

Returns

General-format numeric text with up to 15 significant digits. The result uses fixed notation when appropriate and scientific notation for sufficiently small or large magnitudes.

Behaviour

Ordinary finite values use the Windows system decimal separator and never use a thousands separator. Scientific output uses E and an exponent sign only when required by Velox's general-format rules.

Errors

Normal finite and special Extended values are formatable. Allocation or locale-initialisation exceptions can propagate; no fallback value is supplied.

Usage notes

Use an explicit mask when an integration contract requires a fixed decimal point, exact scale or prohibited exponent notation. General-format output is best treated as human-readable, locale-sensitive numeric text.

Additional Technical Info​

vxFloatToStr converts an Extended value to Delphi general-format text with 15 significant digits and a fresh Windows system-locale settings record. It is the implementation behind both the script names vxFloatToStr and FloatToStr.

The fictional example is source-reviewed and not executed by the documentation workflow. Its decimal character follows the host system locale.

Implementation

uPSI_vxFormats registers vxFloatToStr directly against vxFormats.vxFloatToStr; it also maps the unprefixed FloatToStr name to that same address. Each call creates TFormatSettings with LOCALE_SYSTEM_DEFAULT and selects System.SysUtils.FloatToStr(Value, AFormatSettings).

The installed Delphi 37.0 overload calls FloatToText with fvExtended, ffGeneral, precision 15 and zero minimum exponent digits. The RTL removes trailing zeros, emits a decimal separator only when needed, and chooses fixed or scientific notation from the value's exponent.

Edge cases and quirks

  • The terminal emits NAN, INF and -INF for the corresponding special floating values.
  • Fifteen significant output digits can lose information from an 80-bit Extended. On targets where Extended aliases Double, the input itself has Double precision; scripts should not assume identical least-significant digits across build targets.
  • General formatting chooses fixed or scientific notation from the decimal exponent and the 15-digit precision. In the installed RTL implementation, exponents greater than 15 or less than -3 select scientific notation; callers should not treat this as a fixed-decimal report formatter.
  • Negative zero and floating rounding follow the installed Delphi RTL and target floating environment.
  • Locale is the Windows system default, not a user's display preference and not Delphi's mutable global FormatSettings. Output can differ between Velox hosts.
  • The unprefixed FloatToStr is an exact scripting alias in the current registration. There is no semantic reason to call one name over the other.

Side effects

The call creates local settings and allocates the result string but does not mutate script or product state. The vxFormats unit has a separate initialisation action that sets the initialising thread locale and refreshes Delphi globals once.

Performance and concurrency

The RTL uses a small fixed conversion buffer, so cost is effectively constant for the bounded floating representation. Explicit local settings avoid races on Delphi's global format record; a system-locale change can affect a subsequent call.

Related entries

  • FloatToStr is registered to this exact implementation.
  • vxStrToFloat parses the corresponding system-locale decimal/scientific grammar.
  • vxCurrToStr formats fixed-point Currency without reducing it to floating point.

External references

Created 2026-07-15