Add RunEndBool encoding for efficient boolean array compression

[joseph-isaacs]

Summary

This PR introduces a new RunEndBool encoding specialized for compressing boolean arrays with long runs. Since boolean values strictly alternate in run-end encoding, we can store only the run end positions plus a single start flag, rather than maintaining a separate values array like the generic RunEnd encoding.

Changes

1. New vortex-runend-bool crate with:

   • RunEndBoolArray type and metadata serialization
   • Specialized compression/decompression for boolean runs
   • Compute kernels: filter, take, invert, is_constant, is_sorted, min_max
   • Scalar access and slicing operations
   • Efficient run-preserving filtering that merges adjacent runs with identical values

2. Shared run-end indexing logic extracted to encodings/runend/src/shared.rs:

   • RunEndIndex trait for common "ends" indexing operations
   • find_physical_index() and find_slice_end_index() free functions
   • validate_ends() validation helper
   • Both RunEnd and RunEndBool now implement this trait

3. Integration with compression pipeline:

   • New BoolRunEndScheme in vortex-btrblocks for automatic bool array compression
   • Estimates compression ratio based on run count (profitable when average run length > 8)
   • Registered in the default compression scheme list

4. File format support:

   • RunEndBool arrays serialize/deserialize correctly in Vortex file format
   • Added round-trip test in vortex-file

Design Notes

• Boolean runs strictly alternate, so run i has value value_at_index(i, start) where even indices equal start and odd indices equal !start
• The encoding preserves run structure through filtering and slicing operations
• Sparse filters (< 10% of runs kept) decode via per-index binary search; dense filters scan runs once
• All compute operations respect optional validity arrays

[codspeed-hq]

Merging this PR will not alter performance

⚡ 3 improved benchmarks
❌ 4 regressed benchmarks
✅ 457 untouched benchmarks
⏩ 1091 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	decompress_rd[f64, (100000, 0.0)]	845.5 µs	1,025.2 µs	-17.53%
❌	Simulation	decompress_rd[f32, (100000, 0.0)]	499.1 µs	587.5 µs	-15.04%
❌	Simulation	bitwise_not_vortex_buffer_mut[128]	186.1 ns	215.3 ns	-13.55%
❌	Simulation	bitwise_not_vortex_buffer_mut[1024]	246.4 ns	275.6 ns	-10.58%
⚡	Simulation	null_count_run_end[(10000, 4, 0.01)]	125.4 µs	91.4 µs	+37.2%
⚡	Simulation	decompress_rd[f64, (100000, 0.01)]	981.2 µs	846.5 µs	+15.92%
⚡	Simulation	decompress_rd[f64, (100000, 0.1)]	981.2 µs	846.6 µs	+15.9%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing claude/runend-bool-compression-r148yw (6046f48) with develop (679e2c5)²}

Footnotes

1. 1091 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. ↩

2. No successful run was found on develop (9f494a1) during the generation of this report, so 679e2c5 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩