Release history
GoldenMatch releases
All releases
150 shown
No immediate action
goldengraph-native-v0.1.0
Maintenance
Routine maintenance and dependency updates.
No immediate action
goldenmatch-native-v0.1.19
Maintenance
Routine maintenance and dependency updates.
No immediate action
goldenmatch-native-v0.1.18
Maintenance
Routine maintenance and dependency updates.
No immediate action
golden-suite-v0.3.0
Maintenance
Routine maintenance and dependency updates.
No immediate action
goldenanalysis-v0.4.0
Maintenance
Routine maintenance and dependency updates.
goldencheck-v3.0.0
Breaking risk
⚠ Upgrade required
- If previously relying on Polars for default scanning, install the `[polars]` extra or switch to Arrow‑native APIs.
- For performance‑critical workloads, consider installing the `[native]` extra to add the compiled kernel accelerator.
Breaking changes
- `scan_file`, `scan_dataframe`, and CLI `check` now operate without Polars, using a pyarrow.Table scan frame and fused Rust/Arrow kernels (fallback to pyarrow.compute when native kernel absent).
- `pyarrow` is promoted from an optional to a BASE dependency; `polars` is removed from the base install and moved to the `[baseline]` extra for scipy stat/drift/correlation features and the `[polars]` extra for `scan_dataframe(pl.DataFrame)` convenience overload.
- `inferred_type` output changes from raw Polars dtype strings to a neutral vocabulary (str/int/uint/float/date/datetime/bool/other).
Notable features
- `pyarrow` becomes a required base dependency enabling Arrow‑native scanning.
- Optional extras `[baseline]` and `[polars]` provide scipy statistics/drift/correlation (using Polars) and DataFrame convenience overloads respectively.
Full changelog
The Flip: the default scan path is now Arrow-native and Polars-free.
Breaking
scan_file/scan_dataframe/ the CLIcheckrun WITHOUT Polars. The scan frame is a pyarrow.Table; profiling routes through the fused Rust/Arrow kernels (with a pyarrow.compute fallback when the native kernel is absent).pyarrowis now a BASE dependency.polarsis NOT: it moved to the[baseline]extra (the opt-in scipy stat/drift/correlation subsystems still use it) and the[polars]extra (thescan_dataframe(pl.DataFrame)convenience overload).inferred_typenow emits a neutral dtype vocabulary (str/int/uint/float/date/datetime/bool/other) instead of raw Polars dtype strings.- Sampling is now an owned deterministic sample (was the Polars-PRNG
sample(seed=42)), stable across runs and workers.
Contract
Non-stat findings are byte-identical to the 2.x Polars path (verified by a finding-set differential: strict Jaccard 1.000 across a 9-dataset corpus, 0 kernel-bug divergences). The only intended differences are the neutral dtype strings and the owned sample.
Install
pip install goldencheck-- Arrow-native scanning, no Polars.pip install goldencheck[native]-- adds the compiled kernel accelerator.pip install goldencheck[baseline]-- adds the scipy + Polars stat/drift/correlation features.
See packages/python/goldencheck/CHANGELOG.md for details.
v3.0.0
Breaking risk
⚠ Upgrade required
- Add migration conversion e.g., `pl.from_arrow(result.golden)` where Polars APIs were previously used.
- New env variables for fused routing: GOLDENMATCH_GOLDEN_FUSED, GOLDENMATCH_MATCH_FUSED, GOLDENMATCH_FUSED_PRESSURE_FRACTION, GOLDENMATCH_FUSED_RSS_SCALE, GOLDENMATCH_FUSED_BYTES_PER_PAIR, GOLDENMATCH_FUSED_BYTES_PER_CELL, GOLDENMATCH_FUSED_BLOCK_CONCURRENCY.
Breaking changes
- DedupeResult.golden / .dupes / .unique and MatchResult.matched / .unmatched now return pyarrow.Table instead of polars.DataFrame (Polars-eviction W5). Migration requires a one‑liner conversion e.g., pl.from_arrow(result.golden).
Notable features
- Experimental GOLDENMATCH_FRAME=arrow file ingest lane with Polars‑parity reader corpus and frame‑backend harness.
- Fused auto‑routing at golden and match seams: default live routing of covered slow‑path runs to Arrow‑native kernels (golden_fused) achieving ~2× lower peak RSS, controlled via env knobs GOLDENMATCH_GOLDEN_FUSED, GOLDENMATCH_MATCH_FUSED, etc.
- Standalone fused golden‑record kernel (golden_fused) providing byte‑identical output at reduced memory usage; requires goldenmatch-native 0.1.13.
Full changelog
Changed
- BREAKING: result frames are now
pyarrow.Table.DedupeResult.golden/
.dupes/.uniqueandMatchResult.matched/.unmatchedreturn
pa.Tableinstead ofpl.DataFrame(Polars-eviction W5, spec
docs/superpowers/specs/2026-07-09-goldenmatch-polars-eviction-design.md).
Migration is a one-liner:pl.from_arrow(result.golden). Inputs are
unchanged (polars/pandas/arrow all accepted).to_csvand the notebook
_repr_html_render from Arrow natively. - The arrow ingest lane runs the expression stages (row-ids, standardize,
exact matchkeys) eagerly on the Frame seam; validation rules and auto-fix
are seam-routed with probed arrow twins (RE2 regex owned contract).
Added
- Experimental GOLDENMATCH_FRAME=arrow lane: file ingest via pyarrow with a polars-parity reader corpus and a frame-backend differential harness (Polars-eviction W1).
- Fused auto-routing at the golden + match seams (controller-driven). The pipeline now auto-routes covered slow-path runs to the fused Arrow-native kernels, byte-identically, with no config change and per-surface kill-switches. Golden routing is the broad, LIVE win: at the golden seam the pipeline tries
golden_fused.run_golden_fused_arrowby default on every covered slow-path config and reaches golden output at roughly 2x lower peak RSS; it declines to the classic builder (unchanged output) when the config is uncovered, the native kernel is absent, fullClusterProvenancelineage is requested, or the run is fast-path-eligible. Surfaced onDedupeResult.golden_fused_used; kill-switchGOLDENMATCH_GOLDEN_FUSED=0. Match routing is a documented DORMANT capacity-survival scaffold, not a live default. A controller post-step (maybe_route_fused_match) can short-circuit the match stage to the fused match kernel under an estimated-peak-RSS pressure gate, but the gate is deliberately narrow (auto_split=False, no identity/adaptive/memory/llm_boost/confidence_majority/full-provenance, not across-files-only, a covered weighted matchkey, and real memory pressure). Because zero-config auto-config commitsauto_split=Trueand explicit configs bypass the controller, it effectively never fires by default. When it does fire it is a capacity mode that shedsscored_pairs, cluster-confidence, and lineage to survive; markedDedupeResult.match_fused_capacity_modeand controller-telemetryrule_name+fused_match_post_step. Kill-switchGOLDENMATCH_MATCH_FUSED=0; the wake-up path (a postflight-threshold fast-follow, or an explicit opt-in) is a follow-up. New env knobs (core/fused_routing.py,ExecutionPlan.use_fused_match):GOLDENMATCH_GOLDEN_FUSED,GOLDENMATCH_MATCH_FUSED,GOLDENMATCH_FUSED_PRESSURE_FRACTION(default 0.65), and the est-RSS calibration coefficientsGOLDENMATCH_FUSED_RSS_SCALE/_BYTES_PER_PAIR/_BYTES_PER_CELL/_BLOCK_CONCURRENCY. No new native symbols (reuses thegolden_fused/match_fusedkernels), no new MCP tools / CLI commands / A2A skills. - Fused golden-record kernel (
golden_fused) - standalone, Arrow-native. A new one-FFI-call survivorship kernel (goldenmatch.core.golden_fused.run_golden_fused_arrow, gated bygolden_fused_ready) builds golden records byte-identically to the classicbuild_golden_records_batchpath at ~2x lower peak RSS (measured viascripts/bench_golden_fused_memcap.py+ thebench-golden-fused-memcapworkflow). It covers every Rust-portable survivorship rule in a single native call: scalar strategies (majority / unanimous / first_non_null / longest / source_priority / most_recent / most_complete), quality-weight tie-breaks,field_groupscorrelated survivorship, conditionalfield_rules(predicate AST lowered to a kernel RPN IR viagolden_fused_predicate.lower_predicate/predicate_lowerable),cluster_overrides, andconfidence_majority, plus per-fieldsource_row_idprovenance. Configs that need row-level validators, Python plugins, or LLM survivorship are declined (golden_fused_readyreturns False) and fall through to the classic path unchanged. Not yet wired into the pipeline - this ships as a standalone, benched kernel (composability + peak-RSS win), so no existing user-facing behavior changes. Requiresgoldenmatch-native0.1.13 (the depended-ongolden_fusedsymbol ships in that wheel); without a republished wheel,pip install goldenmatch[native]users transparently degrade to the classic golden path (not a correctness bug - byte-identical output either way). - MCP naming aliases for cross-language parity. The Python and TypeScript MCP servers previously exposed the same operations under different names (
find_duplicates/dedupe,match_record/match,explain_match/explain_pair,profile_data/profile, plus TS'sexplain_cluster). Both servers now answer to both names via non-breaking aliases, so an agent trained against either server can call the other. The Python server gainsdedupe/match/explain_pair/profile/explain_cluster; the TypeScript server gainsfind_duplicates/match_record/explain_match/profile_data. The API-parity gate enforces the nine names staysharedinparity/goldenmatch.yaml. Aliases are excluded from thegoldensuite-mcpaggregated surface so the suite'sprofilestill resolves to goldencheck's file-profiler.
Config change
goldencheck-v2.0.0
Breaking risk
Breaking upgrade
Dependencies
Polars optional for CSV/full scans
goldenflow-v2.0.0
Breaking
⚠ Upgrade required
- If you use transform_df(pl.DataFrame), read_file, or any Polars objects, migrate by installing with `pip install goldenflow[polars]`. Prefer the new Polars‑free API `transform(data, config)`.
Breaking changes
- Polars is no longer a base dependency; it has been moved to the [polars] extra. Pip install goldenflow now runs without Polars by default.
Full changelog
BREAKING: Polars is no longer a base dependency. pip install goldenflow now runs Polars-free by default on goldenflow-native (a base dep); Polars moves to the [polars] extra. All 113 transforms + CSV/Parquet/Excel/DB read + zero-config run without Polars. Migration: pip install goldenflow[polars] if you use transform_df(pl.DataFrame) / read_file / Polars objects; prefer the Polars-free transform(data, config). See CHANGELOG 2.0.0.
No immediate action
golden-suite-v0.1.3
Breaking risk
goldenflow & goldenflow-native version bumps
No immediate action
goldensuite-mcp-v0.3.0
Maintenance
Routine maintenance and dependency updates.
v2.0.0
Breaking risk
⚠ Upgrade required
- Run the identity migration (`goldenmatch identity migrate-ids --path` or `--dsn`) before upgrading to v2.
- No longer set `GOLDENMATCH_CLUSTER_FRAMES_OUT`; clustering will use Arrow frames‑out automatically.
Breaking changes
- Removed Identity `:hash:` scheme; `GOLDENMATCH_IDENTITY_ID_SCHEME` flag and legacy lookup candidate are gone. Persisted DBs with `:hash:` keys will split on next run; must run `goldenmatch identity migrate-ids --path` or `--dsn` before upgrading.
- Removed `GOLDENMATCH_CLUSTER_FRAMES_OUT` environment variable. Arrow frames-out path is now the only clustering path (output‑equivalent). Public `build_clusters` retained as a frames‑backed adapter.
- Removed `RunHistory.cheapest_healthy()` method; replace with `pick_committed()`.
Notable features
- Stable Sail IdentityGraph API added.
- Identity migration command `migrate-ids` introduced.
- Fellegi‑Sunter block scoring performance improved (~3.5x faster on tiny‑block shapes).
Full changelog
GoldenMatch 2.0.0 -- first backwards-incompatible major.
This release removes four deprecation-window items that shipped with a
1.x runway. The pipeline behavior is output-equivalent; the breaks are in
removed escape hatches, internal shims, and the legacy identity id scheme.
BREAKING CHANGES
- Identity
:hash:scheme removed. The legacy:hash:lookup candidate
andGOLDENMATCH_IDENTITY_ID_SCHEMEare gone. A persisted identity DB
still holding:hash:-keyed records will SPLIT on the next run.
Rungoldenmatch identity migrate-ids --path <db>(or--dsn) BEFORE
upgrading. Un-fingerprintable rows keep their:hash:id. GOLDENMATCH_CLUSTER_FRAMES_OUTremoved. The Arrow frames-out path is
the only clustering path now (output-equivalent). Publicbuild_clusters
is preserved as a frames-backed adapter.RunHistory.cheapest_healthy()removed -- usepick_committed()._scale_aware_backend(internal) removed -- backend selection routes
through the v3 planner.
Migration
If you persist an identity DB, run the migration before upgrading:
goldenmatch identity migrate-ids --path <db> # or --dsn <postgres-dsn>
goldenmatch identity migrate-ids --path <db> --dry-run # counts only
Full guide: https://docs.bensevern.dev/ (Guides -> Migrating to v2).
See packages/python/goldenmatch/CHANGELOG.md for the complete 2.0.0 notes
(Added: stable Sail IdentityGraph API, identity migrate-ids; Performance:
Fellegi-Sunter block scoring ~3.5x faster on tiny-block shapes).
No immediate action
goldenmatch-duckdb-v0.6.0
New feature
Native-direct SQL graph UDFs + embedding
No immediate action
goldencheck-types-js-v0.1.0
Maintenance
Routine maintenance and dependency updates.
No immediate action
goldencheck-js-v0.4.0
Maintenance
Routine maintenance and dependency updates.
Review required
goldenmatch-js-v0.3.0
Breaking risk
Breaking upgrade
Auto‑config, checks, classifiers, options, edge
No immediate action
v1.4.0
New feature
Quality scoring, data‑driven strategies, LLM auto‑enable