perf benchmarking suite

[OmarTawfik]

Frameworks

• Cargo:
  • https://github.com/bheisler/iai
  • https://github.com/bheisler/criterion.rs
  • https://crates.io/crates/inferno
  • https://crates.io/crates/deepsize
• NPM:
  • https://www.npmjs.com/package/benchmark
  • https://www.npmjs.com/package/tinybench

Todo

• Implement a few core user cases (examples below).
• Implement a script to run these tests locally and collect results across different commits: infra measure test_name --ref $HASH1 --ref $HASH2 --ref $HASH3.
• Measure CPU load, memory usage, and resulting binary/package size.
• Can we run in CI? report via mermaid graphs?

Scenarios

1. Parsing files: small vs large contracts.
2. Parsing files: flat types vs nested expressions.
3. Using cursors to select specific kinds.
4. Using cursors to navigate up/down/around the tree.
5. Running queries E2E with/without captures.
6. Inspecting every node in the tree, which is critical for things like formatters, which will eventually need to access all of them.