GoldenMatch

vgoldenmatch-native-v0.1.3 scope: goldenmatch-native Bugfix

This release fixes issues for SREs watching stability and regressions.

Published 1mo Data Pipelines & ETL

View tool

✓ No known CVEs patched

Read the diff → Tool health → What is this tool? →

✓ No known CVEs patched in this version

Topics

data-cleaning data-engineering data-matching data-quality deduplication entity-resolution

+14 more

fellegi-sunter fuzzy-matching knowledge-graph llm master-data-management mcp-server polars pprl python record-linkage rust splink typescript zero-config

Summary

AI summary

Fixed near‑zero progress in native block‑scoring kernel, reducing ~190 s jobs to ~1.2 s.

Changes in this release

Type	Severity	Summary	CVE
Performance	Medium	Improves small/medium native block-scoring calls by executing in calling thread, avoiding rayon overhead; 286.9s -> 1.2s (240x faster). Improves small/medium native block-scoring calls by executing in calling thread, avoiding rayon overhead; 286.9s -> 1.2s (240x faster). Source: llm_adapter@2026-06-04 Confidence: low	—
Bugfix	Medium	Fixes thread parking in native block-scoring kernel causing ~190s delay on Linux AMD EPYC. Fixes thread parking in native block-scoring kernel causing ~190s delay on Linux AMD EPYC. Source: llm_adapter@2026-06-04 Confidence: low	—
Bugfix	Low	Updates `goldenmatch_native.__version__` to read from installed dist metadata instead of hardcoded string. Updates `goldenmatch_native.__version__` to read from installed dist metadata instead of hardcoded string. Source: llm_adapter@2026-06-04 Confidence: high	—

Full changelog

Fixes the real root cause of issue #688: the native block-scoring kernel's internal rayon collect parked the calling thread on a LockLatch futex that made near-zero forward progress on some Linux runners (8-core AMD EPYC), turning a sub-second scoring job into ~190s.

The kernel now scores small/medium calls in the calling thread (no rayon, no latch) and only fans out to rayon above a candidate-pair threshold (tunable via GOLDENMATCH_NATIVE_RAYON_MIN_PAIRS). The Python caller already parallelizes across buckets, so cross-bucket parallelism is unchanged.

Verified on the ubuntu-latest-xlarge runner that originally hit the wedge: 286.9s -> 1.2s dedupe wall (240x), output byte-identical.

Also: goldenmatch_native.__version__ now reads from installed dist metadata (was a hardcoded string that reported stale on the 0.1.2 wheel).

See #688 and #692.

View diff on GitHub

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

Share this release

Share on X Share on Bluesky

Track GoldenMatch

Get notified when new releases ship.

About GoldenMatch

All releases →

Related context

Related tools

vector

Earlier breaking changes

v3.1.0 `GOLDENMATCH_FRAME=polars` now requires the `[polars]` extra; raises error without it.
vgoldencheck-v3.0.0 `inferred_type` emits neutral dtype vocabulary (str/int/uint/float/date/datetime/bool/other) instead of raw Polars dtypes.
vgoldencheck-v3.0.0 'inferred_type' now emits a neutral dtype vocabulary instead of raw Polars dtype strings.
vgoldencheck-v3.0.0 `scan_file`, `scan_dataframe`, and CLI `check` now run without Polars, using Arrow-native pyarrow.Table.
v3.0.0 Result frames now return pyarrow.Table instead of Polars DataFrame.