docs: update incremental benchmarks (3.4.0)

[github-actions]

Automated incremental benchmark update for 3.4.0 from workflow run #462.

[greptile-apps]

Greptile Summary

This is an automated benchmark update that appends 3.4.0 measurement data to generated/benchmarks/INCREMENTAL-BENCHMARKS.md, covering the summary table, the "Latest results" section, and the embedded JSON data block.\n\nI verified internal consistency across all three sections:\n\n- Native full build: fullBuildMs: 2276 → displayed as 2.3s ✓; ↓4% vs 3.3.1's 2369 ms ✓\n- Native 1-file rebuild: 448ms; ↑29% vs 3.3.1's 346 ms ✓\n- WASM full build: fullBuildMs: 5033 → 5.0s; ↑15% vs 3.3.1's 4363 ms ✓\n- WASM 1-file rebuild: 617ms; ↑21% vs 3.3.1's 508 ms ✓\n- Speedup ratio: jsFallbackMs: 11.8 / nativeBatchMs: 3.8 ≈ 3.1× ✓\n\nAll percentage deltas, rounded display values, and JSON raw values are consistent. The regression in 1-file rebuild latency and WASM full build that was flagged in a previous review thread has already been acknowledged and tracked in #635 — no further action is required here.

Confidence Score: 5/5

Safe to merge — purely a docs update with correct, internally consistent benchmark data.

Single auto-generated docs file; no code changes. All numeric values cross-check correctly between the summary table, Latest results section, and embedded JSON block. The only known concern (1-file rebuild regression) was already acknowledged in the prior thread and is being tracked separately in #635.

No files require special attention.

Important Files Changed

Filename	Overview
generated/benchmarks/INCREMENTAL-BENCHMARKS.md	Automated addition of 3.4.0 benchmark entries to the summary table, "Latest results" section, and embedded JSON data block; all numbers are internally consistent.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[Release workflow publish.yml] -->|workflow_dispatch| B[Run benchmarks on own codebase]
    B --> C{Engine}
    C -->|Native Rust| D[Full build / No-op / 1-file timings]
    C -->|WASM| E[Full build / No-op / 1-file timings]
    D --> F[Compute import resolution stats]
    E --> F
    F --> G[Calculate % deltas vs previous version]
    G --> H[Update summary table row]
    G --> I[Update Latest results section]
    G --> J[Append JSON entry to embedded data block]
    H --> K[INCREMENTAL-BENCHMARKS.md]
    I --> K
    J --> K
    K --> L[PR #627 opened by github-actions bot]

Loading

_{Reviews (2): Last reviewed commit: "Merge branch 'main' into benchmark/incre..." | Re-trigger Greptile}

[greptile-apps]

Disproportionate 1-file rebuild regression

The 1-file rebuild time increased +29% for native (346ms → 448ms) and +21% for WASM (508ms → 617ms), but the file count only grew by ~7% (442 → 473 files). For a single-file incremental rebuild the total file count should have minimal impact on latency, so this regression is disproportionate to the codebase growth.

The WASM full build also regressed +15% (4.4s → 5.0s), widening the native/WASM gap from ~1.85× to ~2.21×.

These regressions are faithfully recorded here and the numbers are correct — this is just a heads-up that the underlying performance delta in 3.4.0 may be worth investigating before the next release.

[carlos-alm]

Acknowledged — the 1-file rebuild regression (+29% native, +21% WASM) is disproportionate to the ~7% file count growth and worth investigating. Since this is an auto-generated benchmark recording and the numbers are correct, the fix belongs in the codebase (likely in the incremental rebuild path), not in this PR. Will track separately.

[carlos-alm]

Tracked in #635 for follow-up investigation. The regression is real and worth looking into, but as you noted, the benchmark data here is correct — the fix belongs in the incremental rebuild path, not in this docs PR.

[carlos-alm]

@greptileai