Sign and numeric precision

Sign and precision are where formatted numbers stop being an aesthetic choice and start being a correctness one — a truncated price or a missing minus sign is a bug, not a style preference.

Sign

Spec	42	-42	Meaning
- (default)	42	-42	Sign only on negative numbers
+	+42	-42	Always show a sign
(space)	42	-42	Space reserved for positive numbers, aligning with negatives

The alternate form

The # flag changes the presentation of the value itself, not just its padding:

• Integers get their base prefix: 0x/0X for hex, 0b/0B for binary, 0o for octal.
• Floats always show a decimal point, even with zero fractional digits ({:#.0f} on 5.0 gives 5., not 5).

Zero padding

{:08} zero-pads to a width of 8, and the padding is sign-aware — the zeros go between the sign and the first digit, not before it.

Zero-padding and an explicit fill/align cannot both apply — the explicit one wins

fmt::format("{:08}", -42);      // "-0000042" — zero padding, sign-aware
fmt::format("{:*>8}", -42);     // "*****-42" — explicit fill/align, zeros not implied

0 is shorthand for "zero-fill, sign-aware pad." Writing an explicit fill character and alignment overrides it entirely rather than combining with it.

Precision on floats

.N with f, e, or g controls decimal places differently per presentation:

Spec	3.14159
{:.2f}	3.14
{:.2e}	3.14e0
{:.2g}	3.1
{}	3.14159 (shortest round-trip)

Precision on strings

For strings, .N truncates to at most N characters instead of controlling decimal places.

A precision spec bounds a string in a fixed column

fmt::format("{:.5}", "a very long string");  // "a ver"

Combine with a width to build a fixed-width, always-truncated column — see Alignment, fill and width for the layout half.

Default float output

fmt's default float output round-trips exactly — printf's %g does not

With no explicit precision, fmt prints the shortest decimal representation that reads back to the exact same double. printf's %g truncates to a fixed number of significant digits by default and can lose precision silently.

Special values

nan, inf, and -inf print as those literal words. The sign spec still applies to inf: {:+} on positive infinity prints +inf. nan has no sign to apply regardless of the sign spec.

See also

• Type-specific presentation — the presentation types precision and sign apply to.
• Format spec syntax — where sign and precision sit in the full grammar.
• Numeric grouping and locales — the L flag that sits right after precision.
• fmt overview — the full doc set.