#1.4. Identifiers

An identifier is a token that names a variable, function, type, module, etc.

#1.4.1. Format

Identifiers are the default definition of Unicode Standard Annex #31 (UAX #31) (XID_Start / XID_Continue) plus the following: first, the underscore _; second, the mathematical notation profile; third, primes; fourth, emoji.

identifier          ::= identifier_start identifier_continue*
identifier_start    ::= xid_start | "_" | math_start | emoji | flag
identifier_continue ::= xid_continue | math_continue | prime | emoji | emoji_component

#Unicode version

The properties XID_Start, XID_Continue, ID_Compat_Math_Start, ID_Compat_Math_Continue and Extended_Pictographic follow the tables of Unicode 18.0.0 (UCD 18.0.0).

#Mathematical notation profile

This follows UAX #31 §7.1 "Mathematical Compatibility Notation Profile". math_start is ID_Compat_Math_Start plus the n-ary operator symbols ∑ ∏ ∫, and math_continue is ID_Compat_Math_Continue plus the same three symbols. The table below is an informative example (as of UCD 18.0); the properties are authoritative.

∑ ∏ ∫ are added by rok so that they can be used as function names; for this reason they cannot be operator tokens.

#Primes

′ ″ ‴ ⁗ (U+2032-2034, U+2057) may appear only in identifier_continue (f′, x″).

#Emoji

• emoji: code points with the Extended_Pictographic property, excluding those at or below U+00FF (©, ®, etc.). They may appear in both start and continue positions.

• emoji_component: may appear only in continue positions.

  • ZWJ (U+200D) - only between two pictographs
  • Variation selectors U+FE0E / U+FE0F - removed by normalization (1.4.2)
  • Skin tone modifiers U+1F3FB-1F3FF
  • Regional indicators U+1F1E6-1F1FF (flags)
  • Tags U+E0020-E007F

• flag: a flag made of two regional indicators (🇰🇷) may also appear at the start of an identifier.

A keycap sequence (1️⃣) starts with a digit and is therefore not an identifier.

#x² is a name

x² is not x squared but a single name, x². Squaring is written x * x or x.pow(2).

#Valid examples

count
my_var
α
β2
Δt
σ²
x₀
f′
∂u
∇φ
∑
변수1
データ
🚀count
_internal

#Invalid examples

2x         // first character is a digit
my-var     // '-' is not an identifier character
my var     // whitespace
′x         // a prime cannot be the first character
1️⃣         // keycap - starts with a digit

#1.4.2. Normalization

Identifiers are compared after NFC (Normalization Form C) normalization followed by removal of the variation selectors U+FE0E / U+FE0F. The precomposed é and the combining form e + ◌́ are therefore the same identifier, as are ❤️ and ❤. The compiler applies this normalization immediately after tokenizing an identifier.

This decision ensures searchability and consistency of diagnostic messages.

#Confusable warnings

The compiler emits a warning (not an error) in the following cases.

• When distinct names within the same package (regardless of file) have the same skeleton. The skeleton is NFKC normalization plus confusable folding; an implementation may use only a subset of the UTS #39 confusable table. Examples: σ² and σ2; Latin a and Cyrillic а.
• When Cyrillic characters are mixed with Latin or Greek characters within one identifier. Latin-Greek mixing (Δt, σx) is common in mathematical names and is not warned about; neither is mixing Hangul, Han characters or Kana with Latin.

#1.4.3. Case

Identifiers are case-sensitive. count and Count are distinct tokens.

The recommended naming conventions snake_case, PascalCase and SCREAMING_SNAKE belong to the style guide and are not enforced by the specification.

#1.4.4. Collision with keywords

A token that exactly matches a reserved word defined in Keywords is classified as a keyword, not an identifier. Contextual keywords are identifiers outside their designated positions, and after ., in field name declarations, and in named argument positions, even reserved words are identifiers (1.5.5). Raw identifiers that would allow reserved words to be used as identifiers, such as Rust's r#fn, are not adopted; the cost of confusion was judged to outweigh their value.

#1.4.5. Length limit

This specification sets no upper bound on identifier length and follows the general rule of Source files.

#1.4.6. Single underscore

A single underscore (_) is accepted as an identifier but has a special meaning in pattern matching and value discarding. That meaning is defined in the pattern specification. At the lexical stage it is tokenized as a one-character identifier.

Edit this page on GitHub