Depending on the use-case, decimal numbers may be represented as integers or as floating-point numbers with a dot character as the separator for the fractional part (typically one or two digits long). Leading zeroes may be present. Leading (negative) sign characters are possible, although use-cases for them are rare (if any).

Spaces or commas must not be used inside the numeric values.

Scientific notation (with mantissa/exponent) must not be used to represent numeric values set into variables, to serve the values as exact as we have them and keep the client-side parsing simple and predictable.

For example: "01200.2" and "1200.20" are valid, while "1,200.20" and "1200,20" and "1 200.20" and "1.2e4" are invalid.

Programming note: floating-point numbers should be emitted using the %f format specifier in the C printf family of methods and derived methods (including NUT dstate_setinfo() in the driver code). Specifiers like %e and %g which can emit the scientific notation should be avoided when setting variable values (directly in code or when providing format string patterns in mapping tables). They may however be used in debug traces, where reasonable.

Note that in some cases (e.g. USB vendor and product identifiers) technically numeric values may be reported as hexadecimal and should be treated generally as opaque strings (with the consumer ascribing a known meaning to certain variable names).