# FIX Antenna .NET: bounded capture with the original shared runner

Both hosts complete 41 original-SDK pairs, fourteen actual CLI refusals and
492 alterations of actual output copies. Independent collections each execute
once and reproduce every proof. Per host, all 3,408 application frames are
counted, while 2,224 are captured. The complete 1,274 primary spans reproduce
their direct scalar clocks; the original shared runner retains 924 measured
primary and 780 measured scheduler-delay values and excludes 350 warmup/prewarm
samples. No pair or collector is repeated, no predecessor is rebuilt and no
timing, Stage 5 admission or result cell is exported.

## Trace equivalence and bounded normal behavior

Twenty-three traced pairs repeat the qualified original phase recipes using
the new standalone runtime. Complete wire bodies, prepared application bytes,
callback/extraction clocks, ownership, original sequencing, actual deadlines
and warmup/measured partitions pass the original independent phase reader.
The immutable earlier producer remains separate and unchanged.

Eighteen normal pairs cover all four flows with zero/five warmup, repeated runs,
unpaced mode and changed-symbol/touch mixed/tick controls. They use 24 measured
iterations so that every normal case exercises uncaptured application work.
Only the first twelve measured cycles of each run are captured, together with
every original administration frame. Warmup frames do not enter that fixture.
All raw SDK sequence/ordinal and application counts still run on every frame;
continuous sequences and the sparse bilateral fixture membership are checked.
Normal mode writes neither prepared snapshots nor per-cycle string clock traces.

Complete typed extraction remains active on every application message, including
every market-data entry and all tick reaction fields. Independent replay checks
the whole-run sink and every phase sink/count prefix. Preallocated numeric arrays
retain actual deadlines, task starts, prepared starts, callback/primary ends and
completed extraction ends. They are written after measurement. Exact clock
differences reproduce all primary values, actual-minus-due reproduces measured
scheduler delays, and the original result statistics and rate windows are checked.

Original SDK sequence, SendingTime and CheckSum work remains inside public send.
The declared tick boundary is prepared application bytes before SendMessage to
callback entry; OnMessageSent remains after-write capture and never supplies t0.
This qualification cannot promote the boundary to completed-wire transport entry.

## Immutable inputs, source checks and operations

All seventeen SDK/package inputs and twelve compiled payload files remain
unchanged on each host. The original fixed-scale helper and sixteen-file shared
harness DLL are reused directly. Native qualification compiles none of them.
Loaded SDK/helper/harness paths and hashes, actual .NET tool hash, runtime argv,
role masks, measurement-thread affinity and full artifact inventories are bound.
Each host independently verifies its already qualified full-trace phase parent.

The combined original SDK/runtime source suite passes 233 tests without selected
skips. Packaging and affected generation controls pass 64 tests; the installed
payload/profile layer passes 51. Actual launcher tests refuse malformed inputs
before .NET execution. Both independent host reviews are required before any
new immutable bundle can be built or installed.

An initial compile checkpoint precedes the additional direct scalar end-clock
series. A later wrapper stops before compilation because of an incorrect parent
directory suffix. The following compile succeeds with one nullable warning; the
final v4 adds the checked-session annotation and compiles without warnings/errors,
producing the same executable DLL. Original logs remain retained. None of these
compile checkpoints launches a native pair. No unchanged successful native matrix
is rerun for a source/reader correction.

Jobs use normal UID/GID 1000 and functional isolated CPU leases, with HP high
priority and SCAN normal priority. SCAN waits for the pre-existing isolated job;
no other session's job is interrupted. Source builds/tests may use available
shared CPUs concurrently. Both native services exit zero without a main process.
Independent collections run under the BEELINK2 shared lease, once per host.
Generic root helpers, registration, SDK/predecessors and existing results stay
unchanged. Installed-runtime/fixed-entry DAC qualification and fresh quiet
physical report admission remain separate. The .NET matched comparison and full
plan remain incomplete. Website delivery keeps compact summaries and excludes
full per-result CDF payloads.

## Pins and receipts

Native source: `91c1c6eaa9dc43598aa73e8d611ce82c145ab6c8`.
SDK Core: `11b2934a4dd07106292f6d67bf37b65151540efb5e5d2cb0c36bc228d0279f40`.
Fixed-scale helper: `f027e4aea85438d57ca1e7080444254f94d5f80cea2cfe69bff2191344c9ad20`.
Original harness: `5ef1c68bfa62b6bf5fe7f92efd5ec6bff4b91b172b46c937e1c7483fd7cbad5e`.
New runtime: `5bd9b3a9ebab4549ce2acb9a40b6968ed60c294cb9af0c1fd43e22cfe58005d4`.
Final source: `66463d199b04d396f2c004a4e161ac0e54c83fd28e6042bf6fbade8302d8ef1f`.
Final compile inputs: `43d14b06159d2f40a5c27a9ef462d5e472ba64ea6dc9eb95afd0ff57579db6c7`.
Native root: `~/build/codex-fullbench/libhft/fixantenna-runtime-qualification-20261026-v1`.
Independent collections: `~/build/codex-fullbench/libhft/fixantenna-runtime-review-20261026-v1/{hp,scan}`.

| Host / receipt | Value or SHA-256 |
| --- | --- |
| HP administration frames | 412 |
| HP native-receipt.json | `3f0d6586128f019d890eaff68bff02f42ad3c7d4d351b0cf99a6e011d31a4ea5` |
| HP cold-review.json | `4a3912368c8a5530774af5e202532c5984080f10b1f7b5f34a6c3ec12b024003` |
| HP reader-controls.json | `2fa297892f4fc7dc374d7c72b7275d2eb0112f673f83ba0fd6419856e7e1a0d5` |
| HP collection-review.json | `390b4a3208b69ae873e57ceb4b8e1b40153ba6a81a4de4cf69697750b6273e7c` |
| SCAN administration frames | 410 |
| SCAN native-receipt.json | `1cd537728dd84e9c19dd8cd84d6a7685ab8bdd8f6c4adef008d57fcf81a04af6` |
| SCAN cold-review.json | `6bc02e7df36673da4a366ff714d36db9a0d370ff244a584e8fcae76c5278780b` |
| SCAN reader-controls.json | `2fa297892f4fc7dc374d7c72b7275d2eb0112f673f83ba0fd6419856e7e1a0d5` |
| SCAN collection-review.json | `719de9677181d8430b336cc02607fbaec0eee157112d3cbbe6cc3a05a11fb634` |