[OptimistOtaku]

Frame settings resolve background, camera, and cursor styles with three separate scans of the same style segments. Playback and export pay for all three even in gaps where no style is active. With 200 style segments, the committed release benchmark reduces an inactive lookup from 2.169 µs to 0.770 µs (2.82× faster).

Select the winning segment for each group in one pass, then apply only those overrides. Track priority, later start, and later original index still decide each group independently. Numeric comparison preserves tied signed-zero starts. Inactive/invalid segments and non-finite query times remain excluded, and frames without an applicable override still borrow the base configuration. The existing single-group lookup remains available for cursor-only and camera-padding queries.

Benchmark

Windows x86_64, Ryzen 5 5600H, Rust 1.88.0 GNU, repository release profile (opt-level=s, thin LTO). The benchmark compares the previous style_at implementation with the new method in the same binary. Seven alternating paired samples after warmups; medians shown. Each sample uses 10,000 inactive queries or 1,000 active queries. Complete configurations are checked for equivalence before timing, and owned results are dropped within the measurement.

Styles / query	Before	After	Speedup
0 / inactive	0.086 µs	0.078 µs	1.11×
20 / inactive	0.404 µs	0.179 µs	2.25×
200 / inactive	2.169 µs	0.770 µs	2.82×
1,000 / inactive	11.377 µs	3.279 µs	3.47×
20 / active	4.432 µs	3.725 µs	1.19×
200 / active	26.449 µs	25.250 µs	1.05×
1,000 / active	144.583 µs	145.828 µs	0.99×

Active styles still clone the project configuration, so the gain is concentrated in inactive queries; the largest active control was approximately 1% slower. This measures style resolution in the shared frame-settings path, not overall rendering or FPS.

cargo test -p cap-project --locked --release --lib benchmark_style_resolution -- --ignored --nocapture

Validation

• cargo fmt --all
• cargo check -p cap-project --locked
• cargo test -p cap-project --locked --lib: 167 passed, three benchmarks ignored by default; the style benchmark was run separately.
• cargo clippy -p cap-project --all-targets --locked -- -D warnings
• git diff --check

The differential test compares resolved configurations and borrowed/owned behavior against the original implementation, including overlapping tracks, independent group winners, tied starts, signed zero, disabled/invalid intervals, exact boundaries, and NaN/infinite query times. Existing style precedence, padding, migration, and serialization tests pass. The native GPUI app was not launched locally.

This branch is based directly on upstream main and is independent of #2956.

Confidence Score: 5/5

This PR appears safe to merge; no actionable issues were found.

Summary

style_at now selects background, camera, and cursor overrides in one scan instead of three.

• Frame settings find background, camera, and cursor styles in one pass.

_{Reviews (1) · Last reviewed commit: "perf(project): resolve style groups in o..." · Reviewed by Greptile}

[vercel]

@OptimistOtaku is attempting to deploy a commit to the Cap Software Inc Team on Vercel.

A member of the Team first needs to authorize it.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

An organization owner can choose the repository organization as the billing source for reviews requested by members with a paid Copilot license. Learn about Copilot code review billing.