[OVVO-Financial]

Automatic parity-cache refresh from the R source of truth.

• R repository: OVVO-Financial/NNS
• R commit: 7b07a837c130ba7220dbe16c584221508757d5ae
• NNS version: 14.0
• Source package: NNS_14.0.tar.gz
• Windows binary verified: NNS_14.0.zip
• Live regeneration result: failure
• Cache-only verification result: failure

The exact R source package was installed and the committed parity cache was regenerated. sync/nns_source.json records this R commit as the behavioral-truth provenance. Any remaining Python/R parity mismatches are retained in the workflow diagnostics and should be repaired against this R-authored baseline.

Cache changes by function

function	before	after	added	removed	changed
Co.UPM_nD	6	6	0	0	2
LPM.VaR	18	18	0	0	4
LPM	90	90	0	0	1
NNS.ANOVA.custom	15	15	0	0	4
NNS.ARMA	13	13	0	0	10
NNS.ARMA.optim.custom	2	2	0	0	2
NNS.ARMA.pred_int	2	2	0	0	2
NNS.CDF.custom	25	24	0	1	4
NNS.M.reg	20	20	0	0	19
NNS.MC.stat_summary	1	1	0	0	1
NNS.SD.cluster.dendrogram	2	2	0	0	2
NNS.SD.cluster	7	7	0	0	2
NNS.VAR.custom	10	9	0	1	9
NNS.VAR	12	12	0	0	11
NNS.boost.factor_predictor	2	2	0	0	2
NNS.boost.multi_factor_predictor.positional.v1	2	2	0	0	2
NNS.boost.numeric	30	30	0	0	28
NNS.caus	18	18	0	0	18
NNS.copula	10	10	0	0	6
NNS.dep	32	32	0	0	30
NNS.diff.custom	5	5	0	0	3
NNS.distance.bulk.custom	12	12	0	0	7
NNS.distance	21	21	0	0	3
NNS.gravity	6	6	0	0	2
NNS.meboot.stat_summary	1	1	0	0	1
NNS.mode	24	24	0	0	9
NNS.moments	5	5	0	0	1
NNS.norm	8	8	0	0	1
NNS.part	73	73	0	0	49
NNS.reg.factor_dimred	3	3	0	0	3
NNS.reg.factor_predictor.v2	1	1	0	0	1
NNS.reg	295	277	0	18	158
NNS.seas	11	11	0	0	8
NNS.stack.factor_predictor.v2	3	3	0	0	3
NNS.stack.mixed_factor_predictor.v1	2	2	0	0	2
NNS.stack.numeric	37	33	0	4	33
UPM.VaR	18	18	0	0	4
UPM	89	89	0	0	1
dy.d.scalar	7	7	0	0	7
dy.dx.numeric	2	2	0	0	2
dy.dx.overall	1	1	0	0	1

[OVVO-Financial]

R 14.0 cache audit (2026-11-06): This generated reference snapshot is valuable, but do not merge it into main as parity-green. The generation workflow overall concluded success only because its two test phases are continue-on-error; PR metadata explicitly reports Live regeneration result: failure and Cache-only verification result: failure. On the cache-only matrix, Python 3.11 reported 461 failed, 1,279 passed, 12 skipped; live-R generation had 466 failed, 1,283 passed, 3 skipped. The cache-diff table also shows 24 dropped oracle entries (18 NNS.reg, 4 NNS.stack.numeric, 1 NNS.CDF.custom, 1 NNS.VAR.custom). We staged all 59 regenerated shards, while retaining the accepted newer Python parity helper and NNS-core provenance, on top of the cumulative 41-PR implementation stack in draft PR #191: #191 . That is the correct target for reconciliation and a fresh all-entry live-R rebuild. Make the final regen-workflow step fail when either steps.regenerate.outcome or steps.verify.outcome is not success, while preserving diagnostics and the generated PR for investigation.

[OVVO-Financial]

Dependency notice: This was the initial R14 cache PR targeting main; do not merge it separately. Main does not yet contain the 41 stacked Python port PRs, so a cache-only merge would introduce a red parity baseline. The complete source-preserving cache (2,745 entries / 78 shards) and port code now coexist in draft #192 atop #191 / #189. This PR remains open as the provenance record; no work is being deleted or silently closed.

[github-actions]

Superseded by the complete NNS R 14.0 integration in #192, merged as bd19c3303222af46da79d8c97f258248eb5674d2. This initial cache-only PR was NOT merged unchanged. Its snapshot is preserved at archive/r14/pr-190-20261107 (bdd4708246e73fb4d3816be516928964f26aa3f2). The complete source-authored cache and aligned Python implementations landed together after the full Python 3.11-3.14 GitHub matrix passed.