Toolchain development

<!-- Part of the Carbon Language project, under the Apache License v2.0 with LLVM Exceptions. See /LICENSE for license information. SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception -->

Toolchain structure

• Under toolchain/:
  • base/: Base infrastructure and common utilities.
  • check/: Semantic analysis (SemIR generation).
  • lex/: Lexing (Source -> Tokens).
  • lower/: Lowering to LLVM IR.
  • parse/: Parsing (Token -> Parse Tree).
  • sem_ir/: Semantic Intermediate Representation (SemIR) definitions.

Toolchain architecture

• Documentation: Refer to toolchain/docs for detailed architecture design and patterns.
  • Refer to Toolchain Idioms for a comprehensive list of patterns (for example, ValueStore, formatting .def files, struct reflection) used throughout the implementation.
• Builtin Functions: Refer to the Builtin functions skill (SKILL.md) for guidelines on registering, mapping, constant evaluating, and lowering compiler builtin primitives (e.g. "int.convert_float").
• Language server: Refer to the Language server skill (SKILL.md) before working on toolchain/language_server/ or utils/vscode/. Neither follows the patterns described here.
• Phases: Lex -> Parse -> Check -> Lower.
• Definitions: Many kinds (tokens, parse nodes, SemIR instructions) are defined in .def files and expanded by way of macros.
• Handlers:
  • Parser: Handle<StateName> in parse/handle_*.cpp.
  • Checker: HandleParseNode in check/handle_*.cpp.
  • Lowering: HandleInst in lower/handle_*.cpp.
• Iteration: Prefer iterative algorithms over recursive ones to prevent stack exhaustion on complex codebases.

Essential commands

• Test everything: bazelisk test //...
• Test specific target: bazelisk test //toolchain/testing:file_test
• Test specific file: bazelisk test //toolchain/testing:file_test --test_arg=--file_tests=<path_to_carbon_file>
• Build toolchain: bazelisk build //toolchain/...

Updating test data

Carbon tests often use file_test (for example, //toolchain/testing/file_test). For detailed guidelines on authoring tests, including file splits, naming conventions (fail_, todo_), and generating minimal output with SemIR dumps, please refer to the Toolchain tests skill.

If you change compiler behavior, you likely need to update expected test outputs. Do not manually edit thousands of lines of expected output. Use the script:

./toolchain/autoupdate_testdata.py
# Or for a specific file:
./toolchain/autoupdate_testdata.py toolchain/check/testdata/my_test.carbon

Debugging and diagnostics

• Compiler Diagnostics: Refer to the Diagnostics skill (SKILL.md) for strict rules on declaring, formatting, emitting, testing, and styling compiler diagnostic messages (errors, warnings, notes).
• Printing to stderr: Use llvm::errs() << "debug info\n";.
  • Avoid std::cout (it may interfere with tool output).
• SemIR Stringification:
  • SemIR objects often have a Print method or operator<<.
  • inst.Print(llvm::errs())
• Debugging Crashes:
  • Bazel sandboxing can hide artifacts. Use --sandbox_debug if needed, but often running the binary directly from bazel-bin/ is easier for debugging.

Error handling

• No exceptions: Do not use C++ exceptions.
• ErrorOr<T>: Return ErrorOr<T> for fallible operations.
  • Check with if (auto result = Function(); result) { Use(*result); }
• llvm::Expected<T>: Similar to ErrorOr, used when interfacing with LLVM.

Context-Aware Diagnostics

When declaring and emitting errors, ensure semantic wording matches the exact context:

• Semantic Precision: Do not reference "types" when raising errors for unsized expressions like IntLiteral or FloatLiteral. For example, use RealLiteralTooLargeForUnsizedInt instead of a diagnostic referencing an "integer type".
• Wording Consistency: Before declaring a new diagnostic in kind.def, search for existing diagnostics in the targeted implementation files (for example, other uses of MaxIntWidth) to align message structures and parameter expectations.

Casting (LLVM style)

• Use llvm::cast<T>(obj) (checked, asserts on failure).
• Use llvm::dyn_cast<T>(obj) (returns null on failure).
• Use llvm::isa<T>(obj) (boolean check).
• Avoid dynamic_cast and standard RTTI.

Leverage LLVM APIs

Before implementing custom algorithms for mathematical, logical, or bitwise operations, inspect target LLVM ADT class APIs:

• Builtin APIs: Verify if LLVM classes (such as APInt, APFloat, or APSInt) already offer native equivalents (for example, .pow(), ilogb(), .changeSign(), convertFromAPInt()). Avoid duplicate, naive, or inefficient custom loops.

Data structures

• Prefer APIs in common/ and toolchain/base/ over LLVM ADTs. For example, use Map instead of llvm::DenseMap.
• If no Carbon API exists, prefer LLVM ADTs over standard library ones (for example llvm::SmallVector, llvm::StringRef).
• StringRef is a view; be careful with lifetimes.

Common pitfalls

1. Legacy explorer references: The explorer prototype has been moved. Ignore references to it in proposals or old docs; focus on toolchain.
2. Manually updating test files: Always check if autoupdate_testdata.py can do it for you.
3. Using std::string unnecessarily: Prefer llvm::StringRef for arguments.
4. Header includes: Use specific include orders (often enforced by clang-format).
5. Parse node order: Semantics processes parse nodes in post-order; ensure your parser transitions support this.
6. Builtin implementation gaps: If adding a primitive builtin function, make sure you address all phases of the lifecycle: macro definition registration, signature validation, compile-time constant evaluation (interpreter), LLVM IR lowering, and prelude modular implementation bindings (avoiding orphan rules). Refer to the Builtin functions skill (SKILL.md) for details.
7. Premature helper abstraction: Avoid extracting tiny helper functions that are called from exactly one place and do not significantly modularize complex code. Prefer inlining directly to keep the implementation compact, readable, and localized.
8. Redundant bounds calculations: Avoid repeating calculations of complex boundary estimations (such as lower and upper bound estimations). Refactor the logic to calculate unified values once, preserving compactness.
9. Trusting stale clangd diagnostics: In-editor diagnostics are only as good as compile_commands.json. If it predates a newly added file, clangd falls back to a default command and reports nonsense, such as missing standard headers or "no member named None". Regenerate it with ./scripts/create_compdb.py, which only takes a few seconds.