[joseph-isaacs]

Summary

Since #10988, develop stores no String Encoding results, because string-bench emitted no ingest records. This PR lets string-bench write its Vortex file results into the existing compression_sizes and compression_times tables. It needs no new table or schema change.

The website side is vortex-data/vortex.dev#152, which lists these rows in their own String Encoding and String Encoding Size groups. Either PR can merge first. If this one lands first, the string charts appear inside the Compression groups until the website deploys.

Changes

• benchmarks/string-bench/src/serialized.rs: adds SerializedResult::v3_records(), built on the existing v3::compression_size_record and v3::compression_time_record helpers. Each (column, encoder) run emits:

  • one compression_size record: the Vortex file size, plus the column's canonical uncompressed size
  • compression_time records: encode for the file write and decode for the scan, each with its median and every run

• Record keys. Every record uses:

  • dataset strings/<column>
  • the encoder (onpair-12, onpair-16, fsst) as dataset_variant
  • format vortex-file-compressed

  The prefix and the non-null variant keep these keys from colliding with compress-bench rows for the same columns (l_comment, ClickBench URL). The compress-bench summaries only use rows paired with Parquet, so string rows can't change them.

• benchmarks/string-bench/src/main.rs: adds an --ingest-jsonl <path> flag. Only the Vortex file suite is recorded; the codec suite is a local diagnostic that CI doesn't run.

• scripts/string-split.py: passes --ingest-jsonl for each run and merges the outputs into results.ingest.jsonl.

• .github/workflows/develop-bench.yml: sets v4_ingest: true for string-bench and drops the comment saying its results aren't stored.

PR runs need no further change. pr-bench-runner.yml already fetches a develop baseline for string-bench from results.ingest.jsonl, so PR comments will compare against develop once develop has rows.

I haven't compiled this or run clippy locally; CI will cover it.

🤖 Generated with Claude Code

https://claude.ai/code/session_01UeniBE69Jeaf1h8ZkLq7QB

[codspeed]

Merging this PR will improve performance by 10.38%

2 warnings

⚠️ Walltime results may be unreliable

Some benchmarks used the Walltime instrument outside CodSpeed Macro Runners, which can make the results inconsistent.

Learn more about Macro Runners

See affected benchmarks

⚠️ 1 benchmark is dominated by system calls

System calls cannot be simulated consistently and are left out of the measure. The reported value understates the real cost. Switch to the Walltime instrument to measure them.

Learn more about system calls

See affected benchmarks

⚡ 1 improved benchmark
✅ 2317 untouched benchmarks

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Walltime	threshold_avx2[80]	3 µs	2.7 µs	+10.38%

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/upbeat-brown-y7naim (b0d522f) with develop (38d8fbc)}