#6 Foundation and Architecture Why Design a New Programming Language - Milestone — Initial Language Specification and

#6 Foundation and Architecture: Why Design a New Programming Language? -> Milestone — Initial Language Specification and Code Examples

After identifying the motivations, evaluating similar languages, and articulating the design philosophy, we arrive at a critical turning point: the first formal milestone in building our new C-style language—a concrete, initial language specification accompanied by representative code examples. This section sets the groundwork for the parsing engine, interpreter, and further compiler stages, all to be implemented in modern C++20/23.

Rather than attempting to deliver a complete language upfront, this milestone focuses on a minimal but functional subset that reflects the design intent, demonstrates language behavior, and validates the foundational choices through real examples. It serves as a practical prototype to test the syntax, type system, memory model, and runtime behavior using C++ as the host language for implementation.

1. Language Subset Goals for the First Milestone

The initial specification is deliberately compact, focused on proving the most essential aspects of the language:

• Syntax familiarity for C/C++ developers

• Strong static typing and early error detection

• Deterministic and explicit variable lifetime

• Function calls and scope rules

• Minimal standard library (print, input, math)

• Compile-time constant evaluation

• Safe handling of optional and result values

By delivering a working interpreter for this subset, we build a strong foundation for advanced features such as modules, traits, generics, and concurrency in later milestones.

2. Core Language Grammar (Minimal Specification)

Here is a simplified Backus-Naur Form (BNF)-like definition of the core syntax:

program      ::= { function | struct }*
function     ::= "fn" identifier "(" [ parameters ] ")" [ "->" type ] block
parameters   ::= parameter { "," parameter }*
parameter    ::= identifier ":" type
block        ::= "{" statement* "}"
statement    ::= variable_decl | assignment | if_stmt | while_stmt | return_stmt | expr_stmt
variable_decl ::= "let" ["mut"] identifier ":" type "=" expression ";"
assignment   ::= identifier "=" expression ";"
return_stmt  ::= "return" expression ";"
if_stmt      ::= "if" expression block [ "else" block ]
while_stmt   ::= "while" expression block
expr_stmt    ::= expression ";"
expression   ::= literal | identifier | call_expr | binary_expr | grouping
call_expr    ::= identifier "(" [ expression { "," expression }* ] ")"
binary_expr  ::= expression binary_op expression
grouping     ::= "(" expression ")"
type         ::= "int" | "float" | "bool" | "string" | identifier
literal      ::= integer | float | string | boolean

This grammar defines the basic structures required to build meaningful programs. The interpreter will tokenize, parse, and execute code written in this subset using C++20/23 features like std::variant, std::visit, and std::monostate to represent and handle expressions and types.

3. Built-in Types and Rules

Primitive Types

• int: 64-bit signed integer

• float: 64-bit floating-point number

• bool: Boolean true/false

• string: UTF-8 encoded string

Composite Types

• Option<T>: Represents a value that may or may not exist

• Result<T, E>: Represents success or failure of operations

Rules

• All variables are immutable unless marked with mut

• No implicit type conversions (e.g., int to float requires explicit cast)

• Arithmetic follows type safety rules; overflow and underflow behavior are defined

• All function parameters are passed by value unless specified otherwise

4. Semantic Rules and Behaviors

• Scope: Variables and functions are scoped to their enclosing block

• Typing: All expressions and function return types must be explicitly typed

• Control Flow: Conditional and loop expressions work with boolean types only

• Error Handling: No exceptions; Option and Result used for recoverable errors

• Ownership: All values follow value semantics; reference types will be added in a later stage

These behaviors will be validated and enforced by the interpreter’s semantic analysis layer, powered by modern C++’s type traits, templates, and strong typing mechanisms.

5. Example: Program Using the Initial Specification

fn factorial(n: int) -> int {
    if n <= 1 {
        return 1;
    }
    return n * factorial(n - 1);
}
fn main() -> int {
    let x: int = 5;
    let y: int = factorial(x);
    print("Factorial: ", y);
    return 0;
}

This code demonstrates:

• Function definition and recursion

• Immutable variable binding

• Typed function parameters and return type

• Print output using minimal standard I/O

• Basic integer arithmetic

6. Interpreter Architecture Preview (C++20/23)

The interpreter implementing this specification will include:

• Tokenizer (Lexer): Converts source code into tokens using std::regex or custom matchers

• Parser: Generates an AST using recursive descent, backed by std::variant nodes

• AST Types:

  • Expression: base class with variants for literals, binary ops, function calls

  • Statement: variants for declaration, assignment, control flow, return

• Type System: Checked statically using template-like rules with concepts and type traits

• Runtime:

  • Evaluation of expressions using std::visit

  • Memory and variable tracking via scoped environments

  • Function call stack using managed frames

C++23 features like constexpr virtual dispatch, std::expected, and structured bindings will help us maintain readable, robust, and efficient interpreter code.

7. Development Plan for Next Phase

This milestone serves as both a deliverable and a foundation. With the initial grammar and runtime working, the following next steps will include:

• Expanding the parser to support user-defined structs

• Adding pattern matching to Option and Result values

• Introducing loop constructs and basic collection types (array, slice)

• Establishing a standard library interface

• Modularizing the interpreter using C++20 modules and build system integration

Conclusion

The initial language specification outlined here is a compact yet expressive subset that embodies the language’s core goals: simplicity, safety, clarity, and deterministic behavior. It is designed for immediate implementation using modern C++ tools and constructs, offering a solid launchpad for rapid iteration and refinement.

This milestone reflects a balance between theoretical design and practical engineering. With carefully selected features and minimal syntax, we now have a functional language core that can evolve into a full-fledged system through a staged development process—driven by C++20/23’s expressive power, compile-time evaluation, and modular infrastructure.