[ty] Speed up include filtering for projects with many literal include patterns by dhimasardinata · Pull Request #25266 · astral-sh/ruff (original) (raw)

MichaReiser changed the title~~[ty] perf: speed up include literal matching~~ [ty] Speed up include filtering for projects with many literal include patterns

May 20, 2026

thejchap pushed a commit to thejchap/ruff that referenced this pull request

May 23, 2026

…e patterns (astral-sh#25266)

Summary

Track literal include globs in a bitset to avoid a linear scan over literal pattern indices.
Keep the existing fast path when an include filter has no literal patterns.
Add a ty benchmark that exercises a 1024-literal include set through Project::is_file_included.

Benchmark

Command:

cargo bench -p ruff_benchmark --bench ty --no-default-features --features ty_instrumented ty_include_filter -- --sample-size 10

Results:

origin/main with this benchmark added: [1.4059 us 1.4082 us 1.4100 us]
This PR: [957.18 ns 961.50 ns 965.61 ns]

This is about 1.46x faster, or a 31.7% reduction in time per literal include match in this benchmark.

Test Plan

cargo test -p ty_project glob::include
cargo check -p ruff_benchmark --benches --no-default-features --features ty_instrumented
cargo bench -p ruff_benchmark --bench ty --no-default-features --features ty_instrumented ty_include_filter -- --sample-size 10

Refs astral-sh#25244

Co-authored-by: Micha Reiser micha@reiser.io

anishgirianish pushed a commit to anishgirianish/ruff that referenced this pull request

May 28, 2026