Skip to main content

vxStrToFloat

function vxStrToFloat(const S: string): Extended;

Example​

procedure ScriptEvent(var Value: variant);
begin
  Value := vxStrToFloat('125');
end;

Usage​

vxStrToFloat converts Windows system-locale numeric text to Extended and raises when parsing fails.

Parameters

NameTypeDescription
Sstring, constOptional sign, digits with an optional system decimal separator, and an optional E/e exponent with its own optional sign and required digits. Leading/trailing ASCII spaces are accepted.

Returns

The parsed Extended. Its actual precision, exponent range and storage follow the Velox build target; on targets where Extended aliases Double, no 80-bit precision is available.

Behaviour

Ordinary input is a complete signed decimal or scientific token. The decimal character must be the Windows system DecimalSeparator; thousands separators and currency symbols are never accepted. Surrounding U+0020 spaces are ignored.

Errors

Malformed syntax, the wrong decimal character and out-of-range input raise Velox EConvertError into the script. Use vxStrToFloatDef or vxTryStrToFloat when failure is expected.

Usage notes

For machine interchange, define and normalise a fixed decimal grammar before calling a system-locale routine. Do not rely on a workstation's display format matching the Velox host.

Additional Technical Info​

vxStrToFloat parses complete decimal/scientific text as Extended using a fresh snapshot of the Windows system locale. It raises when the text is malformed or outside the target floating range. In the current scripting import, StrToFloat is an exact alias of this implementation.

The integer example avoids a locale-specific separator. It is fictional, source-reviewed and not executed by the documentation workflow.

Implementation

uPSI_vxFormats registers both vxStrToFloat and the unprefixed StrToFloat name against vxFormats.vxStrToFloat. Each invocation creates TFormatSettings.Create(LOCALE_SYSTEM_DEFAULT) and calls System.SysUtils.StrToFloat(S, AFormatSettings). The installed Delphi 37.0 overload delegates to the target's text-to-Extended terminal and raises EConvertError when it reports failure.

Edge cases and quirks

  • A dot is not portable. It fails as a decimal character when the system decimal separator is a comma.
  • The exponent marker must be followed by at least one exponent digit. Internal spaces, grouping, suffixes and partial tokens fail.
  • The installed Delphi terminal additionally accepts case-insensitive NAN, INF, +INF and -INF as successful Extended values.
  • Overflow or an invalid floating operation fails rather than returning a clamped finite value.
  • Extended differs by target: its precision/range can change between Win32 and targets where it aliases Double.
  • Locale comes from LOCALE_SYSTEM_DEFAULT, not the current user and not Delphi's mutable global FormatSettings record.
  • The unprefixed StrToFloat is deliberately mapped to this same wrapper, so both names have identical locale, special-value and exception behaviour in Velox.
  • Free Pascal's overload is useful language-family context but its supported floating representations and version history do not establish Delphi behaviour.

Side effects

The call changes no persistent data and uses local conversion state. The vxFormats unit has a separate load-time action that sets the initialising thread locale and refreshes Delphi global settings.

Performance and concurrency

Parsing is linear in input length. A fresh settings record adds a system-locale lookup but isolates the terminal from Delphi global-format mutations; subsequent calls can reflect a system-locale change.

Related entries

External references

Created 2026-07-15