[joseph-isaacs]

Summary

3/5 of the V2 scan stack, stacked on #10549. Split out of #10464.

Stack:

1. IO service (feat(file): a coalescing IO service for planning-protocol scans #10548)
2. Planning driver (feat(scan): a planning driver for planners and morsels #10549)
3. V2 scan in vortex-layout (this PR)
4. Wiring into vortex-file, DataFusion and DuckDB
5. Split-boundary search perf

Adds an opt-in scan (VORTEX_SCAN_V2=1) that lowers a file's layout once into a physical plan and runs each split through planners on the planning protocol, executing every plan on develop's pipeline Scan. No engine selects it until the next PR.

Changes

Only vortex-layout changes.

• Planners (scan/planning): announce, pruning, filter and projection planners. The filter planner orders conjuncts adaptively.
• V2 scan (scan/v2): the V2 ScanBuilder and RepeatedScanV2, plan-derived splits, early IO announcements and file pruning.
  • Each stage of a split runs as one pipeline Scan.
  • A split's scans share a ValueCache that is dropped with the split, so dictionary values are never kept for the rest of a scan.
  • Dictionary-value Shares that no plan reads whole are removed, so pushed-down expressions run over encoded values.
• ScanBuilder::into_parts lets the V2 builder copy the default builder's configuration.

Changes needed to rebase onto develop since #10464:

• V2 lowering builds a LegacyStatsPlan for legacy vortex.stats layouts, which the optimizer rewrites into Zoned, matching the merged feat(layout): zoned pruning, shares, and dictionary value reuse in physical plans #10462.
• share.rs keys plans by PlanByPtr in place of the removed PlanRef::addr.

Tests: scan/v2/tests.rs (including legacy-stats and list cases) and scan/planning/filter/tests.rs.

Checks run:

• cargo clippy -p vortex-io -p vortex-scan -p vortex-layout -p vortex-file --all-targets --all-features -- -D warnings on this commit
• cargo test -p vortex-layout --lib, with and without VORTEX_SCAN_V2=1 (on the top of the stack)

🤖 Generated with Claude Code

https://claude.ai/code/session_01Hai26WGsNERxRHXpA3JboE

---

Generated by Claude Code

[codspeed]

Merging this PR will improve performance by 15.28%

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

⚠️ 1 benchmark measured no execution time

Nothing ran under measurement, usually because the compiler removed the code under test. This result is not comparable, so it counts as unchanged.

Preventing compiler optimizations

• bench_compare_primitive[(10000, 2)]

⚠️ 23 benchmarks spent significant time in system calls

System calls cannot be consistently instrumented, so they are not included in the measure, which understates the real cost. Please switch to the Walltime instrument to accurately measure system calls.

Measurement and system calls

• take_indices[(10000, 4096)]
• test_scan_filter_to_arrow_array[int]
• test_repeated_scan_filter[int]
• test_scan[int_str]
• test_scan_filter[int_float]
• test_scan_filter[int_str_float]
• from_vec_drop_vortex[65536]
• zstd_compress_string
• test_scan[int_str_float]
• test_scan_filter[int_str]
• …and 13 more

ℹ️ 40 benchmarks have no flame graph

The flamegraph could not be generated because we failed to resolve the symbols in the profiling trace. This happens when we fail to harvest information about language frames.

• test_scan_scalar_at[int]
• test_scan_scalar_at[int_str]
• test_scan_filter_to_arrow_array[int_float]
• test_scan_filter_to_arrow_array[int]
• test_repeated_scan_filter[int]
• test_scan_scalar_at[int_str_float]
• test_scan_read_all_to_arrow_array[int]
• test_repeated_scan_scalar_at[int_str_float]
• test_scan_to_arrow_array[int]
• test_scan_to_arrow_array[int_float]
• …and 30 more

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 2 improved benchmarks
✅ 2245 untouched benchmarks
⏩ 12 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	WallTime	threshold_neon[80]	4.5 µs	3.8 µs	+16.01%
⚡ ℹ️	Simulation	test_scan_scalar_at[int]	896.1 µs	782.2 µs	+14.56%

Tip

Curious why performance improved? Comment @codspeedbot explain why performance improved on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing ji/scan-v2-3-layout (29f589d) with ji/scan-v2-2-driver (da97422)}

Footnotes

1. 12 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩