# HP completed callback CORE and legacy Rust evidence

The high-priority HP jobs queued before the new helper registration have
completed. Their originals were exported through shared CPU-slot leases;
no other session's work was interrupted or dequeued.

## Callback CORE full local matrix

Twelve actual full local pairs retain 120,000 primary samples, 120,000
scheduler-delay samples and 330,000 application frames per peer. All four
local rate, delay and full stability gates pass. A relocated SDK-free reader
reproduces all twelve original fixture receipts, every raw quantile, the true
maximum extremum and each original timing decision exactly. Original data
and frozen reader bytes/modes remain unchanged.

This is localhost/two-process **S4 diagnostic** evidence. Its new seal contains
178 retained inputs totalling 177,403,338 bytes, with SHA-256
`c21dcb52cb7b5fbfdad675a6dc020be6fe276305fec5a06d50963d2c42adab70`.
It binds the previously qualified callback/default producer and independent
public SDK; it does not award physical or matched-comparison admission.

[Full local originals and independent replay](evidence/cpp-callback-canonical-hp-full-local-20261014/retention-seal.json)
remain separate from the SCAN and physical generations.

## Legacy three-engine physical diagnostics

The earlier TrueFix, Rust Pure and NanoFIX jobs completed 48 actual pairs,
with 360,240 primary and delay samples each. The same unchanged v2 raw/counter
reader used on SCAN checks HP's original file digests, profile/source/runtime
and role identities, commands, namespace/CPU masks, quiet leases, sample
owners/counts, exact quantiles/extrema/rates/delays and physical counter polling.
Every original diagnostic decision is reproduced; original bytes/modes stay
unchanged. This replay adds no build, socket or physical run.

Every row remains **S5 diagnostic** with `comparison_fixture_unaligned`, plus
its original rate/delay failures where present. The legacy wire examples and
sink markers do not establish complete current body/owner proof. Original
contract-audit receipts are hashed without rerunning their auditor. The
reader's seven refusal controls are retained in the separate SCAN record;
this record does not claim fresh HP control copies or canonical admission.

The new HP seal contains 1,575 inputs totalling 20,468,767 bytes, with SHA-256
`a3f627edf7a2bc2237d8abaf6cd8d152c80fb9d9f3868a7528c26e44ae85e7d7`.

[HP legacy originals, frozen producer and independent raw/counter receipt](evidence/legacy-rust-hp-physical-20261014/retention-seal.json)
are retained without borrowing grades from the canonical Rust/NanoFIX comparison.