Gyoji::mir Namespace Reference

Middle intermediate representation (MIR) of a translation unit. More...

Classes
class	Argument
	Function argument. More...
class	BasicBlock
	Basic block of a function. More...
class	Function
	Function inside a translation unit. More...
class	FunctionArgument
	A single named argument to a function. More...
class	Functions
	Container for functions. More...
class	MIR
	The middle-tier intermediate representation (MIR) of a translation unit. More...
class	Operation
	Operations inside basic blocks, the virtual instruction-set of the MIR. More...
class	OperationArrayIndex
	This subclass of OperationBinary represents indexing an array. More...
class	OperationBinary
	This subclass of Operation represents a binary operation. More...
class	OperationCast
	This subclass of OperationUnary represents a cast operation. More...
class	OperationDot
	This subclass of Operation is used to access member variables of classes and other aggregate types by reference. More...
class	OperationFunctionCall
	Function call (invoke) operation. More...
class	OperationJump
	Unconditional Jump. More...
class	OperationJumpConditional
	This operation represents a conditional jump. More...
class	OperationLiteralBool
	Literal bool. More...
class	OperationLiteralChar
	Literal character. More...
class	OperationLiteralFloat
	Literal float. More...
class	OperationLiteralInt
	Literal integer. More...
class	OperationLiteralNull
	Literal null. More...
class	OperationLiteralString
	Literal string. More...
class	OperationLocalDeclare
	Declare a local variable in scope. More...
class	OperationLocalUndeclare
	Un-declare a variable (remove it from scope) More...
class	OperationLocalVariable
	Load a local variable. More...
class	OperationReturn
	Return from a function with a value. More...
class	OperationReturnVoid
	Return from a function without a value. More...
class	OperationSizeofType
	The operation for obtaining the storage size of a type. More...
class	OperationSymbol
	Symbol-table lookup. More...
class	OperationUnary
	This subclass of Operation represents a unary operation. More...
class	Symbol
	A symbol defined globally inside a namespace. More...
class	Symbols
	Symbol table for the intermediate representation (MIR) More...
class	Type
	This represents a type as declared in a translation unit. More...
class	TypeMember
	This represents a typed member variable of a class. More...
class	TypeMethod
	Method of a class. More...
class	Types
	Set of types extracted from a translation unit. More...

Detailed Description

Middle intermediate representation (MIR) of a translation unit.

The types namespace is used to extract and resolve types from the syntax tree.

The MIR is an intermediate representation of a source program. It contains no syntax information and just enough context from the source-code that any errors can be traced back to the source-code for quality error reporting.

The purpose of the MIR is to provide a representation that is at a low enough level that a code-generator such as LLVM can be used to emit assembly, while remaining high-level enough that we can perform analysis and reasoning about the semantics such as implementing a borrow checker.

The MIR consists of two primary components. The first are the Types which represent logical layouts of data. The second are Functions which represent operations that manipulate data (in types).

The purpose of the MIR is to provide a consistent view of the intermediate representation of each of these things together as a unit because many operations like code-generation and analysis rely on them being paired and consistent.

The purpose of this namespace is to process the synax tree and extract the types of objects that will be used later in the compilation and code generation.

This consists of several data types representing the type system as well as a set of classes and functions intended to construct the type information by processing a TranslationUnit.

Type processing may encounter errors where objects are defined incorrectly or types cannot be properly resolved. In these cases, errors may be reported to the error handling system so that the programmer is aware of why the definition or resolution did not succeed.