Skip to content

testing.md

Latest commit

 

History

History
94 lines (55 loc) · 1.99 KB

testing.md

File metadata and controls

94 lines (55 loc) · 1.99 KB

Testing Strategy

Testing Strategy

We employ a pyramid testing strategy to ensure reliability and velocity.

Pyramid:

  1. Unit Tests (Bottom): Fast, isolated, high coverage.
  2. Integration Tests (Middle): Verify component interactions.
  3. E2E Tests (Top): Verify full system behavior via LSP.

Unit Testing

Unit tests should cover all public functions and complex private logic.

Principles:

  • Isolation: Tests must not depend on external systems (filesystem, network).
  • Speed: Tests must run per commit.
  • Coverage: Aim for high branch coverage in core logic.

End-to-End (E2E) Testing

E2E tests verify the system from the user's perspective (LSP client).

Scope:

  • LSP Lifecycle: Initialize, Shutdown, Exit.
  • Diagnostics: Verify reporting of graph errors.
  • Resilience: Verify recovery from invalid states.

Tools:

  • Rust standard test framework with a mock LSP client.

3. Code Coverage Standards

We use cargo-llvm-cov to measure test effectiveness.

Running Coverage Locally:

  1. Install: cargo install cargo-llvm-cov

  2. Add component: rustup component add llvm-tools-preview

  3. Generate report:

    cargo llvm-cov --html
open target/llvm-cov/html/index.html

CI/CD Pipeline:

Every PR includes a Cargo llvm-cov step. We strive for high coverage in core logic.

4. Performance Testing

Derived from: ADR_PERF (Performance Logic)

Strategy:

Performance is critical for recursive graph validation. We use criterion for micro-benchmarks.

  • Location: benches/ directory.
  • Targets: Critical paths in core logic (parsing, linting).

Running Benchmarks:

cargo bench

Goals:

  • Linting: < 50ms for typical workspaces.
  • Graph Generation: Scalable to 1000+ nodes.