Compiler Foundations

Core pipeline

• Lexing -> Parsing -> Semantics -> Lowering -> Runtime
• each stage has a strict contract

Ownership discipline

• lexer: token classes + locations
• parser: grammar + AST shape
• semantics: symbol and type/rule checks
• lowering: representation transform

Why phases help

• each stage has small failure surface
• isolated tests for each transformation
• deterministic errors and artifacts

Tokens, lexemes, AST

• lexeme is raw text
• token is typed chunk + position
• AST is structural meaning

Precedence via grammar

• term/factor hierarchy gives precedence
• loop-based parser with precedence levels keeps associativity stable

Error design

• first-stage failures should be surfaced early
• include line/column + expected token context

Determinism

• same source -> same outputs at every stage
• avoids flaky diagnostics and non-reproducible builds

Practice checks

• map which stage handles which rule
• reason about which stage emits which error class