Available inall subroutines.

Converts the string s to a float value with the given base base. The value base must be a constant integer expression (variables are not allowed).

The following string formats are supported for finite values:

  • Decimal (base 10) floating point syntax. For example, 1.2, -1.2e-3.
  • Hexadecimal (base 16) floating point syntax. For example, 0xA.B, 0xA.Bp-3.

The syntax for these values corresponds to the syntax for VCL FLOAT literals in base 10 and 16 respectively. See VCL Types for details of the FLOAT literal syntax.

Supported bases are 0, 10, or 16.

A base of 0 causes the base to be automatically determined from the string format. In this case, a 0x prefix indicates hex (base 16), and otherwise the base is taken as decimal (base 10).

The syntax is required to match with a corresponding prefix when an explicit base is given. That is, for base 16, the 0x prefix must be present. Likewise, for base 10, the 0x prefix must be absent.

Numbers are parsed with a rounding mode of round to nearest with ties away from zero.

In addition to finite values, the following special string formats are supported:

  • NaN: NaN may be produced by the special format NaN. Note that only one NaN representation is produced.
  • inf, +inf, -inf: Positive and negative infinities may be produced by the special format inf with an optional preceding +/- sign.

The NaN and infinity special formats are case sensitive.

No whitespace is permitted by std.strtof.

On error, fastly.error is set.


If the string could not be parsed, then fastly.error will be set to EPARSENUM.

If the numeric value would cause overflow, then fastly.error will be set to ERANGE.


declare local var.pi FLOAT;
set var.pi = std.strtof(req.http.PI, 10);
if (var.pi >= 3.14 && var.pi <= 3.1416) {
set req.http.X-PI = "Close enough";