# Words

# Table of contents

• Overview
• Keywords
  • Type literals
  • Identifiers
  • Raw identifiers
• Alternatives considered
• References

# Overview

A word is a lexical element formed from a sequence of letters or letter-like characters, such as fn or Foo or Int, optionally preceded by r#.

The exact lexical form of words has not yet been settled. However, Carbon will follow lexical conventions for identifiers based on Unicode Annex #31. TODO: Update this once the precise rules are decided; see the Unicode source files proposal.

Carbon source files, including comments and string literals, are required to be in Unicode Normalization Form C (NFC).

# Keywords

The following words are interpreted as keywords:

• abstract
• adapt
• alias
• and
• as
• auto
• base
• break
• Core
• case
• choice
• class
• constraint
• continue
• default
• destroy
• else
• export
• extend
• final
• fn
• for
• forall
• friend
• if
• impl
• impls
• import
• in
• interface
• let
• library
• like
• match
• namespace
• not
• observe
• or
• override
• package
• partial
• private
• protected
• ref
• require
• return
• returned
• Self
• self
• template
• then
• type
• var
• virtual
• where
• while

# Type literals

A word starting with i, u, or f, followed by a decimal integer, is a _numeric type literal_.

# Identifiers

A word is interpreted as an identifier if it is neither a keyword nor a type literal.

# Raw identifiers

A raw identifier is a word starting with r#. A raw identifier is equivalent to the word following the r# prefix, except that it is always interpreted as an identifier, even if it would otherwise be a keyword or type literal.

Raw identifiers can be used to specify identifiers which have the same spelling as keywords; for example, r#impl. This can be useful when interoperating with C++ code that uses identifiers that are keywords in Carbon, and when migrating between versions of Carbon.

The word doesn't need to be a keyword, in order to support forwards compatibility when a keyword is planned to be added. If word is an identifier, then word and r#word have the same meaning.

# Alternatives considered

Overview:

• Character encoding: We could restrict words to ASCII.
• Normalization form alternatives considered

Type literals:

• Use C++ type keywords with LP64 convention
• Use full type name with length suffix
• Use uppercase for type names
• Support additional bit widths

Raw identifiers:

• Other raw identifier syntaxes
• Restrict raw identifier syntax to current and future keywords
• Don't require syntax for references to raw identifiers
• Don't provide raw identifier syntax

# References

• Proposal #142: Unicode source files
• Proposal #2015: Numeric type literal syntax
• Proposal #3797: Raw identifier syntax