2 Lexical Conventions[Lex]

2.1 Unit of Translation[Lex.Translation]

The text of hlsl programs is collected in source and header files. The distinction between source and header files is social and not technical. An implementation will construct a translation unit from a single source file and any included source or header files referenced via the #include preprocessing directive conforming to the isoC preprocessor specification.

An implementation may implicitly include additional sources as required to expose the hlsl library functionality as defined in (13).

2.2 Phases of Translation[Lex.Phases]

hlsl inherits the phases of translation from isoCPP, with minor alterations, specifically the removal of support for trigraph and digraph sequences. Below is a description of the phases.

1. Source files are characters that are mapped to the basic source character set in an implementation-defined manner.

2. Any sequence of backslash (\) immediately followed by a new line is deleted, resulting in splicing lines together.

3. Tokenization occurs and comments are isolated. If a source file ends in a partial comment or preprocessor token the program is ill-formed and a diagnostic shall be issued. Each comment block shall be treated as a single white-space character.

4. Preprocessing directives are executed, macros are expanded, pragma and other unary operator expressions are executed. Processing of #include directives results in all preceding steps being executed on the resolved file, and can continue recursively. Finally all preprocessing directives are removed from the source.

5. Character and string literal specifiers are converted into the appropriate character set for the execution environment.

6. Adjacent string literal tokens are concatenated.

7. White-space is no longer significant. Syntactic and semantic analysis occurs translating the whole translation unit into an implementation-defined representation.

8. The translation unit is processed to determine required instantiations, the definitions of the required instantiations are located, and the translation and instantiation units are merged. The program is ill-formed if any required instantiation cannot be located or fails during instantiation.

9. External references are resolved, library references linked, and all translation output is collected into a single output.

2.3 Character Sets[Lex.CharSet]

The basic source character set is a subset of the ASCII character set. The table below lists the valid characters and their ASCII values:

An implementation may allow source files to be written in alternate extended character sets as long as that set is a superset of the basic character set. The translation character set is an extended character set or the basic character set as chosen by the implementation.

2.4 Preprocessing Tokens[Lex.PPTokens]

¹

Each preprocessing token that is converted to a token shall have the lexical form of a keyword, an identifier, a constant, a string literal or an operator or punctuator.

Preprocessing tokens are the minimal lexical elements of the language during translation phases 3 through 6 (2.2). Preprocessing tokens can be separated by whitespace in the form of comments, white space characters, or both. White space may appear within a preprocessing token only as part of a header name or between the quotation characters in a character constant or string literal.

Header name preprocessing tokens are only recognized within #include preprocessing directives, __has_include expressions, and implementation-defined locations within #pragma directives. In those contexts, a sequence of characters that could be either a header name or a string literal is recognized as a header name.

2.5 Tokens[Lex.Tokens]

There are five kinds of tokens: identifiers, keywords, literals, and operators or punctuators. All whitespace characters and comments are ignored except as they separate tokens.

2.6 Comments[Lex.Comments]

The characters /* start a comment which terminates with the characters /. The characters // start a comment which terminates at the next new line.

2.7 Header Names[Lex.Headers]

Character sequences in header names are mapped to header files or external source file names in an implementation defined way.

2.8 Preprocessing numbers[Lex.PPNumber]

Preprocessing numbers begin with a digit or period (.), and may be followed by valid identifier characters and floating point literal suffixes (e+, e-, E+, E-, p+, p-, P+, and P-). Preprocessing number tokens lexically include all integer-literal and floating-literal tokens.

Preprocessing numbers do not have types or values. Types and values are assigned to integer-literal, floating-literal, and vector-literal tokens on successful conversion from preprocessing numbers.

A preprocessing number cannot end in a period (.) if the immediate next token is a scalar-element-sequence (2.9.4). In this situation the pp-number token is truncated to end before the period².

2.9 Literals[Lex.Literals]

2.9.1 Literal Classifications[Lex.Literal.Kinds]

2.9.2 Integer Literals[Lex.Literal.Int]

An integer literal is an optional base prefix, a sequence of digits in the appropriate base, and an optional type suffix. An integer literal shall not contain a period or exponent specifier.

The type of an integer literal is the first of the corresponding list in the table below in which its value can be represented³.

If the specified value of an integer literal cannot be represented by any type in the corresponding list, the integer literal has no type and the program is ill-formed.

An implementation may support the integer suffixes ll and ull as equivalent to l and ul respectively.

2.9.3 Floating-point Literals[Lex.Literal.Float]

A floating literal is written either as a fractional-constant with an optional exponent-part and optional floating-suffix, or as an integer digit-sequence with a required exponent-part and optional floating-suffix.

The type of a floating literal is float, unless explicitly specified by a suffix. The suffixes h and H specify half, the suffixes f and F specify float, and the suffixes l and L specify double.⁴ If a value specified in the source is not in the range of representable values for its type, the program is ill-formed.

2.9.4 Vector Literals[Lex.Literal.Vector]

A vector-literal is an integer-literal or floating-point literal followed by a period (.) and a scalar-element-sequence.

A scalar-element-sequence is a vector-swizzle-sequence where only the first vector element accessor is valid (x or r). A scalar-element-sequence is equivalent to a vector splat conversion performed on the integer-literal or floating-literal value (4.10).