- Compiler for the Azoth language
- Currently runs as a tree interpreter
- Goal: Implement basic object-oriented subset similar to C# 2.0 feature set (i.e. including generics)
- The design follows a Rewritable Reference Attribute Grammar (ReRAG) approach
- Tree nodes have attributes which are computed on demand and cached when appropriate
- Attributes can be marked cyclic and will then evaluate cycles to a fixed point
- Code generation is used to make this approach feasible
- *.tree files generate tree node
- *.aspect files add attributes to the tree node and generate partial classes for the methods for those attributes
- Make Azoth source code as correct as possible given current language design
- Ensure language features are documented in language reference
- Implement standard library features as compiler intrinsics initially
- Replace intrinsics with correct standard library code when possible
- Keep namespace hierarchies flat to avoid excessive
usingstatements - Use "Named" instead of "Nominal" for clarity
- Avoid reference cycles in design
- Strong typing throughout the codebase
- Framework: utility code and code that is not specific to the project
- Core: Core compiler functionality including the attribute grammar framework, diagnostics, enums
- Types: Type system implementation
- Symbols: Symbol management (name + type pairing)
- Parsing: The parser which generates a syntax tree
- Syntax: The nodes for the syntax tree
- Semantics: Semantic analysis, including semantic tree nodes
- API: Public interface to the compiler
- CodeGen: Code generation used to generate tree nodes and aspects
Implementing basic object-oriented subset including:
- Stand-alone Functions (no generics)
- Simple Types (int, uint, size, offset)
- Simple Class Declarations
- Fields, Constructors, Basic Methods
- Optional Types
- Basic Control Flow
- Namespaces
- Basic List[T]
- Loop Labels
- Document Comment Validation
- Full Type Inference
Tests are organized in projects named Tests.Unit.*