After Step 7

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
This commit is contained in:
GAJ3T 2026-10-30 13:13:25 -04:00
parent b0b1034b2e
commit 480135539b
10 changed files with 933 additions and 66 deletions

View File

@ -76,10 +76,13 @@ diagnostic operations, explicit source-sensitive local archive save/readback,
strict schema-v1 validation, and shared recording/inspection controls in both
host workflows. Step-07A corrects source-faithful empty/case-variant response
round-trips, v2 archive validation, capture lifetime release, and visible
storage failures. Both worker reports retain their separate evidence; neither
worker completion is orchestrator acceptance. See the [Step-07 worker
storage failures. Step-07B removes the source-close versus callback-unregister
lock cycle while preserving shared disposal completion and failure reporting.
The three worker reports retain separate evidence; none is orchestrator
acceptance. See the [Step-07 worker
report](prompts/Step-07-Report.md) and [Step-07A corrective worker
report](prompts/Step-07A-Report.md).
report](prompts/Step-07A-Report.md) and [Step-07B lock-safety corrective
report](prompts/Step-07B-Report.md).
### Planned

View File

@ -8,8 +8,10 @@ session-owned live dashboard, and the accepted Step-06/06A display-unit
preference. Step-07 implements bounded session recording, local archive
readback and shared host controls. Step-07A corrects archive fidelity and v2
validation, releases stopped capture owners and preserves visible storage
failures. The original Step-07 evidence and Step-07A corrective evidence are
separate; worker acceptance remains pending manual orchestrator review.
failures. Step-07B moves session cleanup outside the lifetime-registration
gate and shares its completion/failure among competing disposers. Each
worker report contains separate evidence; acceptance remains pending manual
orchestrator review.
- [Architecture](architecture.md): actual nine-project graph, responsibilities,
route ownership and provider/reference boundaries.
@ -58,5 +60,6 @@ historical/worker reports for their own execution evidence:
[Step-06 worker report](../prompts/Step-06-Report.md),
[Step-06A corrective worker report](../prompts/Step-06A-Report.md), and
[Step-07 worker report](../prompts/Step-07-Report.md) and
[Step-07A corrective worker report](../prompts/Step-07A-Report.md).
[Step-07A corrective worker report](../prompts/Step-07A-Report.md) and
[Step-07B lock-safety corrective worker report](../prompts/Step-07B-Report.md).
Handoffs are manual; acceptance remains the orchestrator's responsibility.

View File

@ -133,6 +133,12 @@ for inspection/save. UI unmount does not own or stop capture.
Stopped, failed or discarded capture owners are removed from the diagnostic
session's lifetime-resource list; the workspace retains only the current
inspection/save data. An active capture is still drained on source close.
Close marks the session closed, snapshots registrations and publishes the
shared disposal task under the lifetime gate, then releases the gate before
cancellation or resource/control/transport disposal begins. Concurrent
disposers await that same completed task or failure; late registrations are
rejected after close ownership is established. This keeps observer draining
and capture unregistration outside the registration lock.
The Client persists only explicitly saved archives under its app-local
`recordings` directory. Application owns schema-v1 validation and file-storage

View File

@ -580,3 +580,55 @@ environment. See the [Step-07A corrective worker
report](../prompts/Step-07A-Report.md) for the frozen-matrix mapping,
pre/post-fix outputs, self-review and preservation review. Step-08 remains
blocked pending manual review; no Step-08 work began.
## Step-07B lock-safe session disposal verification
The pre-fix reproduction used the real diagnostic protocol/session, capture and
provider transport with a test-only subscription wrapper exposing controlled
dispose barriers. It held a capture callback at limit or observation-error
detach after the capture lock was released, started close/reset until its
second unsubscribe was entered, then released the callback. Both actual tasks
remained incomplete: `Operation completed=False; close/reset completed=False.`
The two focused cases failed as expected after their five-second safety
deadlines (two failed, zero passed); the bounded `dotnet test` host exited
normally after reporting both cases, so no deadlocked process tree required
termination. Command:
`dotnet test .\src\tests\AutoForge.IntegrationTests\AutoForge.IntegrationTests.csproj --configuration Release --filter "FullyQualifiedName~SessionCloseDuringCaptureDetach_DoesNotDeadlockOrLoseFailurePrefix"` (exit 1).
These are deterministic barrier-forced lock-cycle observations, not a
stress-timing claim.
The corrected session publishes one shared disposal completion and snapshots
the lifetime resources under the lifetime gate, then performs cancellation
and resource/control/transport disposal after releasing it. Concurrent callers
observe the same task and cleanup error; late registration rejects once close
ownership has been established. The focused set passed 29/29 recording tests,
including limit-close and error-reset, exact prefix/reason preservation,
replacement-session isolation, concurrent disposal and cleanup failure.
Commands and results:
- `dotnet test .\src\tests\AutoForge.IntegrationTests\AutoForge.IntegrationTests.csproj --configuration Release --filter "FullyQualifiedName~SessionCloseDuringCaptureDetach_DoesNotDeadlockOrLoseFailurePrefix|FullyQualifiedName~ConcurrentSessionDispose_SharesCompletionAndReportsCleanupFailure"` (3 passed)
- `dotnet test .\src\tests\AutoForge.IntegrationTests\AutoForge.IntegrationTests.csproj --configuration Release --filter "FullyQualifiedName~TelemetryRecordingIntegrationTests"` (29 passed)
- `dotnet build .\src\AutoForge.slnx --configuration Release --no-restore` (nine projects; 0 warnings, 0 errors)
- `dotnet test .\src\AutoForge.slnx --configuration Release --no-build` (179 unit + 134 integration; 313 total, zero failed/skipped)
The rebuilt Windows Client was rechecked through its local WebView2 DevTools
endpoint. Static v1 recorded two actual operations (DTC and signal reads),
stopped, closed its source while retaining the unsaved capture, then saved.
After app restart, the archive loaded with its raw bytes and decoded signals.
Dynamics v2 was explicitly stepped to model step 7 / 0.7 seconds, captured four
actual `read.signals` observations with gauges hidden, then stopped and closed
its source while retaining the unsaved capture before saving. A second app
restart loaded both static and dynamics archives with raw data and decoded
source/model context. Metric/Road preferences remained unchanged. The exact
task-owned capture IDs were `3b6bbb098a094ac2ba10c042dcd4fbf0` (static) and
`ac5b4d4e814d4905a0d454292b4dde87` (dynamics); their exact archive files and
task-owned temporary profiles were removed after closing the Client.
The controlled integration tests prove the close/reset lock-cycle resolution;
the native smoke verifies unchanged normal record/stop/close/save/restart/
readback behavior and does not claim native race injection. See the
[Step-07B worker report](../prompts/Step-07B-Report.md) for matrix mapping,
post-green source review and preservation checks. Step-07/07A remain NOT
ACCEPTED pending manual review; Step-08 remains blocked and was not started.

View File

@ -3,15 +3,16 @@
## Status and authority
- Initialized: 2026-10-28.
- Current phase: Step-07A mostly corrected; source-close/capture-unregistration lock ordering remains.
- Active assignment: [Step-07B](Step-07B.md), ready for manual worker execution.
- Last accepted milestone: Step-06 with Step-06A, accepted 2026-10-30.
- Last accepted correction: Step-06A; all Step-06 findings closed.
- Outstanding correction: existing R-03 safe-close invariant; source disposal can deadlock with capture failure unregistration.
- Current phase: corrected recording/archive accepted; saved telemetry replay next.
- Active assignment: [Step-08](Step-08.md), ready for manual worker execution.
- Last accepted milestone: Step-07 with Step-07A/Step-07B, accepted 2026-10-30.
- Last accepted correction: Step-07B; all Step-07 findings closed.
- Outstanding corrections: none.
- Step-05 status: ACCEPTED after Step-05A/Step-05B, 2026-10-29.
- Step-06 status: ACCEPTED after Step-06A, 2026-10-30.
- Step-07 status: REQUIRES FOCUSED CORRECTION / NOT ACCEPTED, reviewed 2026-10-30.
- Step-08 status: recorded telemetry replay; BLOCKED / NOT ASSIGNED until Step-07 acceptance.
- Step-07 status: ACCEPTED after Step-07A/Step-07B, 2026-10-30.
- Step-08 status: READY FOR WORKER; replay controls/clock not implemented yet.
- Step-09 status: calibration/simulated programming; BLOCKED / NOT ASSIGNED until replay acceptance.
- Current state and handoff: [STATE.md](STATE.md).
- Authority: the master charter, repository/handoff amendment, and the owner's
2026-10-28 architectural realignment. The latest decision cancels the dual-shell
@ -375,10 +376,10 @@ Step-05 with Step-05A/05B is accepted: the configured static/dynamic workflow,
single session owner, independent range/quality, safe gauges, decisive tests and
native evidence have been reviewed. Corrected Step-06/06A is accepted: unit
defaults/conversions, rejected selections, unavailable readouts and raw/source
preservation are closed on inspected evidence. Step-07A corrects valid-source
round trips, typed/strict archive validation, normal registration release and
visible storage failures. Source-close overlap still requires Step-07B; that is
the only active assignment. Step-08 replay remains blocked.
preservation are closed on inspected evidence. Corrected Step-07/07A/07B is
accepted: actual bounded capture, source-faithful/strict archives, released
ownership, visible I/O failures and deadlock-free source close are supported.
Step-08 is the only active assignment for saved-data replay, not resimulation.
Dependencies indicate acceptance prerequisites, not permission for automatic
handoff. Later assignments refine these scopes using actual implementation
evidence; they must not silently discard required capabilities.
@ -400,11 +401,11 @@ evidence; they must not silently discard required capabilities.
| 05B | Independent range status and decisive dashboard tests | Remaining 05A R-03/testing contract | ACCEPTED 2026-10-29 | Frozen five-case/five-criterion contract satisfied; reported 134 unit/91 integration passes, clean Release/native recheck and actual pre-fix failure |
| 06 | Imperial-default metric/imperial display units | 05 and corrections accepted; owner request 2026-10-29 | ACCEPTED 2026-10-30 AFTER 06A | Shared persisted Imperial/Road default, exact conversions, source/canonical invariants and corrected rejection/unavailable readouts inspected |
| 06A | Preference rejection and unavailable display readouts | 06 review R-01/R-02 | ACCEPTED 2026-10-30 | Frozen seven-case/five-criterion contract satisfied; reported 179 unit/105 integration passes, clean Release/native evidence and honest binding-event versus browser coverage |
| 07 | Bounded telemetry recording and archive readback | 06 and corrections accepted | REQUIRES FOCUSED CORRECTION / NOT ACCEPTED | Most original scope corrected; concurrent source disposal/capture failure cleanup must not deadlock |
| 07A | Source-faithful archives, bounded lifetime and visible storage failures | 07 review R-01 through R-04 | REVIEWED - FOCUSED FOLLOW-UP REQUIRED | Reported 179 unit/131 integration passes and clean Release/native evidence; archive/I-O fixes supported, existing R-03 close-overlap remains |
| 07B | Deadlock-free source disposal during capture cleanup | Remaining 07A R-03 safe-close contract | READY FOR WORKER | Release lifetime-registration gate before resource/cancellation callbacks; controlled close/limit/failure interleavings with preserved data and single disposal completion |
| 08 | Recorded telemetry replay | 07 and corrections accepted | BLOCKED / NOT ASSIGNED | Play/pause/stop/restart/seek/navigation/speed feed shared presentation from recorded data, with explicit Replay identity and no live reads or physics rerun |
| 09 | Calibration management and simulated programming | 08 | PLANNED | Versioned definitions/diffs, verified baseline and backup, approved simulated write, readback/unrelated-data checks, audit and supported restore; no physical writes |
| 07 | Bounded telemetry recording and archive readback | 06 and corrections accepted | ACCEPTED 2026-10-30 AFTER 07A/07B | Actual bounded observations, source-faithful strict archive save/readback, visible errors, retained capture data and controlled safe-close evidence inspected |
| 07A | Source-faithful archives, bounded lifetime and visible storage failures | 07 review R-01 through R-04 | ACCEPTED 2026-10-30 AFTER 07B | Archive/typed-validation/normal-release/panel fixes and source-close follow-up all closed |
| 07B | Deadlock-free source disposal during capture cleanup | Remaining 07A R-03 safe-close contract | ACCEPTED 2026-10-30 | Shared disposal task published under gate, cleanup outside; actual pre-fix close/reset deadlocks and controlled correction, reported 179 unit/134 integration passes and native baseline |
| 08 | Recorded telemetry replay | 07 and corrections accepted | READY FOR WORKER | Strict saved-archive load, RecordedReplay context, prefix-correct gauges/navigation and one cancellable playback owner; no live calls, resimulation, interpolation or archive mutation |
| 09 | Calibration management and simulated programming | 08 accepted | BLOCKED / NOT ASSIGNED | Versioned definitions/diffs, verified baseline and backup, approved simulated write, readback/unrelated-data checks, audit and supported restore; no physical writes |
| 10 | Local AI, knowledge, and validated runbooks | 09 | PLANNED | Replaceable local inference, applicable knowledge retrieval, structured logical runbook, independent validation/approval and simulation-only execution |
| 11 | Comparative evaluation and complete synthetic demonstration | 10 | PLANNED | Complete demonstration below plus deterministic fault/regression matrix and independent verification of intended and unintended changes |
| 12 | Verified OBDX Pro GT read-only integration | 11 plus documented device research/access | PLANNED / HARDWARE GATE | Source-backed device support matrix, explicit physical target, real read-only identification/diagnostics/telemetry recording where supported; no write path |
@ -811,24 +812,40 @@ history, not repeated redesign without evidence.
Worker reports 179 unit + 131 integration passes (310 total, zero failed/
skipped), clean complete Release build and native static/dynamic save/restart/
readback smoke. The orchestrator ran no application checks.
- Most corrective scope is satisfied. One tightly coupled source-close edge
remains: cleanup starts under the lifetime gate and can wait for an observation
callback that now needs that gate to unregister. Prepared single-issue
- At that review most corrective scope was satisfied, but cleanup under the
lifetime gate could wait for an observation callback needing that gate to
unregister. Prepared single-issue
[Step-07B](Step-07B.md) under existing R-03/C-06, with no archive or replay scope.
- Reviewed and accepted Step-07B against its frozen five-case/four-criterion
contract. Close ownership/resource snapshot/shared completion are published
under the lifetime gate; cancellation and cleanup start outside it.
Actual protocol/subscription barriers cover limit-close and failure-reset,
preserve the one-observation prefix/reason and show replacement isolation.
- Worker reports actual pre-fix blocked-operation/close failures, corrected
concurrent disposal/same-error/count assertions, complete Release build zero
warnings/errors and 179 unit + 134 integration passes (313 total, zero failed/
skipped), plus native stopped-unsaved/source-close/save/restart/readback.
The orchestrator ran no application checks.
- Accepted corrected Step-07/07A/07B with no Step-07C, and prepared
[Step-08](Step-08.md) for a single saved-data playback owner and shared
numeric/bar presentation. Original source/archive bytes and unit behavior stay
immutable; equal timestamps, seek prefixes and cleanup are frozen up front.
### Remaining
- Worker execution of Step-07B, then review the controlled safe-close interleaving
and preserved original/corrective evidence before playback.
- Step-08 replay is blocked and not assigned. Recording/archive exists but remains
unaccepted; display units and earlier milestones remain accepted.
- Worker execution of Step-08, then one consolidated review against its frozen
replay prefix/time/owner/UI matrix and shared quality gate. Correct genuine
unmet requirements as Step-08A before calibration work.
- Step-09 calibration is blocked and not assigned. Replay is assigned work,
not implemented functionality; recording/archive and earlier work are accepted.
- Profile/module metadata, synthetic virtual-module communication and deterministic
dynamics/v2 and corrected dashboard are accepted; later features remain planned.
Preserve the complete sequence/safety gates.
### Known blockers and uncertainties
The single open Step-07 R-03 follow-up is source-close lock ordering:
No blocking Step-07 deficiency remains. Its source-close follow-up below is
historical and closed by accepted Step-07B:
- DiagnosticSession.DisposeAsync holds _lifetimeGate while invoking
DisposeResourcesAsync. Its synchronous prefix can call capture disposal,
@ -840,12 +857,11 @@ The single open Step-07 R-03 follow-up is source-close lock ordering:
needs a controlled competing close/limit-or-error callback, while retaining
shared disposal completion, cleanup failures and frozen capture data.
This is source inspection, not an orchestrator-executed deadlock reproduction.
Step-07B must reproduce the interleaving safely and fix the root lock ordering.
That was source inspection, not an orchestrator-executed reproduction. Step-07B
supplied the actual bounded failing tests and corrected source ordering.
The four original Step-07 findings below are historical. Step-07A has corrected
their inspected archive/I-O/normal-release paths; only the safe-close overlap
above remains:
The four original Step-07 findings below are historical. Step-07A corrected
their archive/I-O/normal-release paths and Step-07B closed the safe-close overlap:
- R-01: archive validation requires response bytes for every ProtocolError, even
though an actual empty reply yields that outcome with null source trace bytes.

View File

@ -5,14 +5,14 @@
| Field | Value |
|---|---|
| Updated | 2026-10-30 |
| Phase | Step-07A mostly corrected; source-close lock-order follow-up remains |
| Active step | Step-07B - Deadlock-free source disposal during capture cleanup |
| Assignment | [Step-07B.md](Step-07B.md) |
| Phase | Corrected recording/archive accepted; saved telemetry replay next |
| Active step | Step-08 - Recorded telemetry replay |
| Assignment | [Step-08.md](Step-08.md) |
| Assignment status | READY FOR WORKER / AWAITING MANUAL HANDOFF |
| Expected worker report | `prompts/Step-07B-Report.md` - not yet created |
| Last accepted correction | Step-06A - accepted 2026-10-30 |
| Last accepted major milestone | Step-06 with Step-06A - accepted 2026-10-30 |
| Outstanding correction | Existing R-03 safe-close invariant: lifetime/observer lock inversion during failure/limit unregistration |
| Expected worker report | `prompts/Step-08-Report.md` - not yet created |
| Last accepted correction | Step-07B - accepted 2026-10-30 |
| Last accepted major milestone | Step-07 with Step-07A/Step-07B - accepted 2026-10-30 |
| Outstanding corrections | None |
| Step-02 status | ACCEPTED after Step-02A |
| Step-03 status | ACCEPTED after Step-03A/Step-03B |
| Step-04 status | ACCEPTED after Step-04A/Step-04B |
@ -20,11 +20,12 @@
| Step-05 status | ACCEPTED after Step-05A/Step-05B |
| Step-05A review | All original corrective criteria closed with Step-05B |
| Step-06 status | ACCEPTED after Step-06A |
| Step-07 status | REQUIRES FOCUSED CORRECTION / NOT ACCEPTED |
| Step-07A review | Archive/I-O/normal release corrected; concurrent source-close edge requires Step-07B |
| Step-08 status | Recorded telemetry replay; BLOCKED until Step-07 acceptance |
| Worker execution | Step-07A returned; Step-07B not invoked |
| Immediate next action | User manually directs the worker to execute Step-07B only; no replay work |
| Step-07 status | ACCEPTED after Step-07A/Step-07B |
| Step-07A review | All original corrective criteria closed with Step-07B |
| Step-08 status | READY FOR WORKER; saved-data replay only |
| Step-09 status | Calibration/simulated programming; BLOCKED until Step-08 acceptance |
| Worker execution | Step-07B returned and accepted; Step-08 not invoked |
| Immediate next action | User manually directs the worker to execute Step-08 and shared quality gate only |
| Next orchestrator action | After manual return, inspect the assignment, report, implementation, tests, Git changes, documentation and recorded command evidence |
## Repository baseline at initialization
@ -928,11 +929,11 @@ no application build/test, native launch or failing reproduction. Preserve the
returned report unchanged. [Step-07A](Step-07A.md) consolidates the four existing
requirements; Step-08 remains blocked and no replay assignment is created.
## Step-07A review and focused safe-close follow-up
## Historical Step-07A review and focused safe-close follow-up
Decision: **MOST SCOPE CORRECTED; FOCUSED FOLLOW-UP REQUIRED**, 2026-10-30.
Step-07/07A remain unaccepted until [Step-07B](Step-07B.md) closes the existing
R-03 safe-close requirement. Do not redo the supported archive/I-O fixes.
Decision at that review: most scope was corrected, but Step-07B was required
for the existing R-03 safe-close overlap. The closing review below now accepts
corrected Step-07/07A/07B without redoing the supported archive/I-O fixes.
Reviewed [Step-07A](Step-07A.md), its [report](Step-07A-Report.md), all changed
source paths and new regression bodies, the shared v2 validator, resource
@ -976,7 +977,42 @@ early. Preserve a single shared disposal completion and aggregate cleanup behavi
The worker's seven added integration cases and 310-case/clean Release/native
evidence remain valid positive reports, distinct from the missing competing-close
proof. All original reports remain unchanged; Step-08 is still blocked.
proof at that review. All original reports remain unchanged; Step-08 was blocked
until the closing review below.
## Step-07B acceptance and corrected Step-07
Decision: **ACCEPTED**, 2026-10-30. Frozen C-01..C-05 and AC-01..AC-04 are
satisfied. Step-07 with Step-07A/07B is accepted; no Step-07C is required.
Reviewed [Step-07B](Step-07B.md), its [report](Step-07B-Report.md), the focused
source disposal change, actual protocol barrier theory and concurrent-disposal
regression, related docs/README and existing Release/native artifacts.
| Criterion | Decision | Evidence |
|---|---|---|
| AC-01 | ACCEPTED | Actual pre-fix close/limit and reset/error lock-cycle failures; shared task/state published under gate and cleanup invoked after release; controlled operations/close/reset finish with exact frozen prefix/reason and isolated replacement |
| AC-02 | ACCEPTED | Concurrent callers get the same task and cleanup exception, controlled resource disposes once, late registration rejects; original resource/control/transport ordering and aggregation preserved |
| AC-03 | ACCEPTED ON WORKER EXECUTION EVIDENCE | Reported complete nine-project Release build zero warnings/errors; 179 unit + 134 integration passes (313 total, zero failed/skipped); native static/dynamic unsaved retention, source close, save/restart/readback |
| AC-04 | ACCEPTED | Actual pre/post and native/control distinction, own post-green source-path review, related docs and protected/history/manual workflow preserved |
The two barrier theory cases and one concurrent-dispose case preserve the prior
310 and yield 313. The five-second deadline is a safety guard after explicit
barriers, not a probabilistic timing race. The pre-fix focused process reported
both tasks incomplete and exited with failures; no native race is claimed.
The helper records genuine cleanup success/failure in the already-published
shared completion. It does not remove unregistration, drop accepted observations,
clear required unsaved data or publish success before disposal.
The orchestrator ran no application build/test/native/workload command and did
not execute the deadlock reproduction. Graph/source/artifacts corroborate worker
execution evidence, not independent runtime verification.
Original Step-07 and both corrective findings are closed. Preserve the 303-/
310-/313-case reports and original archive/raw/source/native evidence.
[Step-08](Step-08.md) is prepared for saved-data replay with explicit
RecordedReplay context, deterministic prefix reduction and one playback owner.
No live source/session, physics execution or archive rewrite is authorized.
## Prerequisites, limitations, and known issues
@ -1004,24 +1040,24 @@ proof. All original reports remain unchanged; Step-08 is still blocked.
Step-04B. Step-05's original and follow-up findings are closed. Preserve the 190-/
202-/207-case reports and original numerical/native evidence distinctly.
- Profile metadata/UI, synthetic diagnostics, deterministic dynamics/v2 and the
corrected dashboard/display units are accepted. Recording/archive exists but
is unaccepted. Replay, calibration values, AI and physical hardware operations
corrected dashboard/display units/recording archive are accepted.
Replay, calibration values, AI and physical hardware operations
remain unimplemented.
- No application build or test has been executed by the orchestrator.
## Review and progression gate
On manual return, evaluate Step-07B and the original Step-07/07A safe-close contract/shared gate using the
On manual return, evaluate Step-08 and its frozen replay contract/shared gate using the
assignment, worker report, actual code/tests, dependency graph, Git changes,
README/docs, and recorded build/test evidence. Do not accept solely on a success
claim; do not claim independent runtime verification.
- Inspect lifetime-gate release before callbacks, controlled close/limit/error
interleavings, shared disposal/error completion and preserved capture state.
Preserve prior archive/I-O/native evidence; no new playback requirements.
- If accepted: close Step-07A/07B, accept corrected Step-07, update roadmap/state, prepare Step-08 replay
- Inspect explicit replay provenance, exact prefix/seek/tie semantics, one clock
owner/drain/generation, source/error projection, shared gauges/units and true
offline behavior. Do not add interpolation/streaming/layout/format requirements.
- If accepted: accept Step-08, update roadmap/state, prepare a bounded Step-09 calibration assignment
and stop for manual handoff.
- If deficient: issue narrowly scoped Step-07C only for a material unmet contract/
- If deficient: issue narrowly scoped Step-08A only for a material unmet contract/
regression, then stop. Planned new features are not corrections.
- If blocked: retain the active step, document the prerequisite and required
worker evidence; do not advance or mark unexecuted checks as successful.
@ -1090,13 +1126,15 @@ claim; do not claim independent runtime verification.
| 2026-10-30 | Step-07A returned and reviewed | Archive/I-O/normal unregistration fixes and reported 310-case/clean Release/native evidence inspected; most corrective scope supported |
| 2026-10-30 | Existing R-03 close-overlap remains | Lifetime gate held during synchronous disposal prefix can wait for observer callback that needs the lifetime gate to unregister |
| 2026-10-30 | Step-07B prepared | Single lock-order/safe-close correction with controlled reproduction and preserved disposal/data semantics; replay remains blocked |
| 2026-10-30 | Step-07B returned and ACCEPTED | Actual pre-fix close/reset deadlocks, corrected gate/task ordering, preserved prefix and shared cleanup-failure assertions plus reported 313-case/clean Release/native evidence inspected |
| 2026-10-30 | Corrected Step-07 ACCEPTED | Original and 07A/07B criteria closed; source archive/limits/storage/default-unit behavior preserved; no 07C |
| 2026-10-30 | Step-08 prepared | Saved validated archive, explicit RecordedReplay context, prefix-correct gauges/seek and one cancellable clock; manual handoff pending, calibration blocked |
## Manual handoff instruction
Provide [Step-07B.md](Step-07B.md), original [Step-07.md](Step-07.md),
[Step-07A.md](Step-07A.md) and the shared
Provide [Step-08.md](Step-08.md) and the shared
[worker gate](WORKER-QUALITY-GATE.md) to the separate worker and request execution
of the focused correction only. The worker must produce `prompts/Step-07B-Report.md`
and stop for review. Do not begin Step-08 replay.
of that assignment only. The worker must produce `prompts/Step-08-Report.md`
and stop for review. Do not begin Step-09 calibration.
Return manually to the orchestrator for review. Reports, including final chat
reports, must follow the single plain triple-backtick copy-window requirement.

139
prompts/Step-07B-Report.md Normal file
View File

@ -0,0 +1,139 @@
```
# Step-07B worker report
## Status
COMPLETE — Step-07B implementation and verification are complete. Manual
orchestrator review remains pending. Step-07 and Step-07A remain NOT ACCEPTED;
Step-08 remains blocked and was not started.
## Implementation
`DiagnosticSession.DisposeAsync` now uses `_lifetimeGate` only to establish
close ownership, publish the closed state, snapshot/clear registered resources
and publish a single shared disposal task. After releasing the gate, one
caller starts cancellation and resource/control/transport cleanup. Concurrent
callers return the same task and observe its actual completion or failure.
Late resource registration remains rejected after close ownership begins.
Existing resource ordering and error aggregation are unchanged.
The test-only `BarrierObservationProtocolClient` forwards actual calls and
observations from `SyntheticDiagnosticProtocolClient` and places a controlled
barrier in the actual capture subscription's dispose path. It exposes no
production race selector and does not mock the session, capture or protocol.
The cleanup-error regression uses a task-owned `IAsyncDisposable` resource.
Changed files:
- [DiagnosticSessionService.cs](../src/AutoForge.Application/Features/Vehicles/Diagnostics/DiagnosticSessionService.cs)
- [TelemetryRecordingIntegrationTests.cs](../src/tests/AutoForge.IntegrationTests/TelemetryRecordingIntegrationTests.cs)
- [README.md](../README.md)
- [docs/README.md](../docs/README.md)
- [architecture.md](../docs/architecture.md)
- [build-and-test.md](../docs/build-and-test.md)
## Pre-fix reproduction
Command:
`dotnet test .\src\tests\AutoForge.IntegrationTests\AutoForge.IntegrationTests.csproj --configuration Release --filter "FullyQualifiedName~SessionCloseDuringCaptureDetach_DoesNotDeadlockOrLoseFailurePrefix"`
Exit 1: the two theory cases failed and zero passed. Each case used controlled
barriers to establish the exact order: an actual diagnostic operation reached
capture limit/error detach after releasing the capture lock; close/reset
entered the second subscription-dispose call while the callback still owned
observation dispatch; then the callback barrier was released. After the
five-second safety deadline the test reported, for both the close and reset
cases, `Operation completed=False; close/reset completed=False.` This is the
observed lock cycle, not an inferred timing race or a compile failure. The
bounded xUnit process exited normally with the failures; no process tree had
to be terminated and no intentionally blocked task remained in a normal
passing run.
## Frozen corrective matrix
| Requirement | Evidence and result |
|---|---|
| C-01 normal stop/close with or without active capture | Existing stop-drain, repeated-registration, source-close retention, stopped saveability and cross-scope tests remain in the focused suite. Native static/dynamics workflow was also repeated. PASSED. |
| C-02 limit-detach competing with source close | `SessionCloseDuringCaptureDetach_DoesNotDeadlockOrLoseFailurePrefix(reset: False)` forces the callback/close interleaving above. Post-fix both operation and close complete, the one-observation prefix remains `Incomplete / LimitExceeded`, and session registration is released. PASSED. |
| C-03 observation-error detach competing with reset | The same regression with `reset: True` forces error detach against `DiagnosticSessionService.ResetAsync`. Post-fix both operation and reset complete; the original one-observation prefix remains `Error / ObservationFailure`; the replacement has a new session identity and its real signal read does not mutate the old snapshot. PASSED. |
| C-04 concurrent disposal, late registration and cleanup failure | `ConcurrentSessionDispose_SharesCompletionAndReportsCleanupFailure` verifies concurrent callers receive the same `Task`, late registration throws `ObjectDisposedException`, the controlled resource is disposed once, and both callers receive the same explicit cleanup exception. Barrier release is in `finally`. PASSED. |
| C-05 existing archive/registration/storage/panel/source/model/unit/native paths | All 310 Step-07A cases remain in the 313-test full suite; the three new theory/test cases pass. Fixed archive and capture limits, source semantics and native record/save/readback are preserved. PASSED. |
| AC-01 C-01..C-03 safe close/reset | Actual pre-fix lock-cycle reproduction and controlled post-fix completion/prefix/reason/replacement assertions. PASSED. |
| AC-02 C-04 shared cleanup semantics | Concurrent disposal identity/count, rejection and shared cleanup error assertions plus source ordering review. PASSED. |
| AC-03 C-05 regressions/build/native | 179 unit + 134 integration tests, nine-project Release build and native static/dynamics smoke. PASSED. |
| AC-04 scope/history/docs/self-review/report | Related documentation updated; protected files/history preserved; this report saved and worker stops before Step-08. PASSED. |
## Post-fix commands and results
Focused, build and full-suite commands ran sequentially:
| Command | Exit | Result |
|---|---:|---|
| `dotnet test .\src\tests\AutoForge.IntegrationTests\AutoForge.IntegrationTests.csproj --configuration Release --filter "FullyQualifiedName~SessionCloseDuringCaptureDetach_DoesNotDeadlockOrLoseFailurePrefix|FullyQualifiedName~ConcurrentSessionDispose_SharesCompletionAndReportsCleanupFailure"` | 0 | Three cases passed, zero failed/skipped. |
| `dotnet test .\src\tests\AutoForge.IntegrationTests\AutoForge.IntegrationTests.csproj --configuration Release --filter "FullyQualifiedName~TelemetryRecordingIntegrationTests"` | 0 | 29 passed, zero failed/skipped. |
| `dotnet build .\src\AutoForge.slnx --configuration Release --no-restore` | 0 | All nine projects, including Windows Client; zero warnings and zero errors. The XAML source-generation notice was informational. |
| `dotnet test .\src\AutoForge.slnx --configuration Release --no-build` | 0 | 179 unit + 134 integration = 313 passed; zero failed/skipped. |
The full suite preserves the 310-case Step-07A baseline (179 unit and 131
integration) and adds the two barrier theory cases plus one concurrent-dispose
case. Existing fixed limits remain 4,096 observations, 16 MiB encoded
observations, 1 MiB header, 32 MiB archive and JSON depth 128.
## Native versus controlled evidence
The rebuilt Windows Client was operated through its local WebView2 DevTools
endpoint:
- Static v1 recorded two actual operations (`read.dtcs` and `read.signals`),
stopped, then closed its source while the unsaved capture remained
inspectable. It was explicitly saved afterward. After app restart, the
archive loaded with its original raw bytes and two decoded signals.
- Dynamics v2 was explicitly stepped to model step 7 / 0.7 seconds. With
gauges hidden, two actual telemetry reads captured four actual
`read.signals` observations at the same model context. The capture was
stopped and the source closed before save; the stopped unsaved data remained
inspectable. After app restart, the archive loaded with raw data and decoded
source/model/configuration/run context.
- Both archives were loaded after restart. Metric/Road preferences remained
unchanged. Exact task-owned IDs were `3b6bbb098a094ac2ba10c042dcd4fbf0`
(static v1, two observations) and `ac5b4d4e814d4905a0d454292b4dde87`
(dynamics v2, four observations). After closing the Client and DevTools
listener, only those exact verified archive files were removed. The app's
existing/default WebView profile and unrelated application data were not
removed or reset.
Native evidence verifies normal static/dynamic record, stop, source close,
retained capture, explicit save, restart and readback. The exact race is
proved by controlled integration barriers, not by a human-timed native claim.
## Documentation, self-review and preservation
Updated the architecture contract to state that close ownership, closed-state
publication, resource snapshot and shared task publication occur under the
lifetime gate, while cancellation and resource/control/transport disposal run
after the gate is released. The root and docs READMEs now describe the
Step-07B correction and link this report. The build guide records the actual
pre-fix reproduction, final commands/counts and native/control distinction.
Step-07 and Step-07A reports and their positive archive/native evidence remain
unchanged.
Post-green self-review traced `DiagnosticSession.DisposeAsync` through close,
reset, cancellation, capture `StopForSourceClose`, limit/error detach,
unregistration, control and transport disposal. The gate is no longer held
while callbacks or observer drain run. `_disposeTask` is published before
another caller can start cleanup; registration checks the closed state under
the same gate; unregistration remains available after the close snapshot.
Disposal error aggregation and completion ordering are preserved. The
barrier tests assert operation and close/reset completion together, the exact
failure prefix/reason, registration release and replacement isolation. The
concurrent cleanup test releases its barrier in `finally`. No additional
issue was found in the review.
`git diff --check` passed. Only the six listed implementation/test/documentation
files and this requested report were changed. Step-07/07A assignment files,
historical reports, Step-07B assignment, state, roadmap and quality gate remain
unchanged. No package/project/system/workload changes, staging, commit, push,
or Step-08 work occurred. Step-07/07A acceptance remains for manual review.
Stop here.
```

332
prompts/Step-08.md Normal file
View File

@ -0,0 +1,332 @@
# Step-08 - Recorded telemetry replay
## STEP IDENTIFIER
Step-08. Corrected Step-07/07A/07B was accepted on 2026-10-30.
This is the sole active implementation assignment. Step-09 calibration is blocked.
## STEP TITLE
Replay validated saved captures through shared numeric/bar presentation with
explicit recorded provenance, deterministic navigation and one playback owner.
## OBJECTIVE
One user outcome: open a saved capture and inspect/replay its recorded signals
without a vehicle session, original profile catalogue or running model.
One primary new contract: a cancellable archive-only playback clock and exact
observation-prefix projection.
Read [STATE.md](STATE.md), [ROADMAP.md](ROADMAP.md),
[archive v1](../docs/telemetry-archive-v1.md), [dashboard contracts](../docs/live-dashboard.md)
and [WORKER-QUALITY-GATE.md](WORKER-QUALITY-GATE.md).
Make the frozen matrix executable before coding, self-review after the first
green run, write `prompts/Step-08-Report.md`, and STOP.
## ARCHITECTURAL CONTEXT
Keep all nine projects, one native Client and pure generic Components.
Application owns replay loading/control/reduction; UI owns a routed feature
and shared presentation. Existing local archive storage and schema-v1 reader
are the only data input. No new project, dependency, transport, protocol, schema,
physics clock, generic timeline/event framework or global active-vehicle registry.
Source/tests remain under `src`, tests under `src/tests`.
Use the existing
[ExecutionTargetKind.RecordedReplay](../src/AutoForge.Core/ExecutionTargets/ExecutionTargetKind.cs:3)
for current replay context, plus a replay-instance identity and capture ID.
Retain the archive's original provider/target/session/profile/config/run as
original metadata, not a current Simulation/PhysicalHardware connection.
Do not fabricate a DiagnosticSession, live provider/feed or verified identity.
## CURRENT IMPLEMENTATION STATE
Accepted features supply strict bounded v1 archives, actual raw observations
with positive-operation-bound decoded signals, complete/incomplete/error captures,
local file list/load, source/diagnostic/model timing and shared unit preferences.
Worker baseline: 179 unit + 134 integration passes (313 total, zero failed/
skipped), clean full Release build and native capture/save/restart/readback.
The orchestrator did not run those commands.
Useful current surfaces:
- [ITelemetryArchiveStore](../src/AutoForge.Application/Features/Vehicles/Telemetry/Recording/ITelemetryArchiveStore.cs)
and [archive models](../src/AutoForge.Application/Features/Vehicles/Telemetry/Recording/TelemetryArchiveModels.cs).
- [Dashboard models](../src/AutoForge.Application/Features/Vehicles/Telemetry/LiveDashboard/LiveDashboardModels.cs)
and [live projection](../src/AutoForge.Application/Features/Vehicles/Telemetry/LiveDashboard/LiveDashboardAcquisition.cs).
- [Display-unit presentation](../src/AutoForge.UI/Features/DisplayUnits/DisplayUnitPresentation.cs),
[generic gauge](../src/AutoForge.Components/LiveDashboardGauge.razor) and
[current panel/editor](../src/AutoForge.UI/Features/Vehicles/Telemetry/LiveDashboard/LiveDashboardPanel.razor).
Replay does not yet exist. Inspect Git/prior art and preserve accepted source,
archive, lifetime, unit and safety behavior. Do not overwrite dirty work/history.
## REQUIRED IMPLEMENTATION
### Frozen contract matrix
| ID | Input/precondition | Required result |
|---|---|---|
| C-01 | Valid saved archive, with original source/profile/model no longer available | Explicit user load produces RecordedReplay context, immutable original metadata and a stopped before-first cursor; no live source operation or autoplay |
| C-02 | Four-observation fixture below | Exact recorded prefix/value/quality/error state at each applied count; recorded time is not inferred from diagnostic/model clocks |
| C-03 | Time seek between records, backward seek, Next/Previous, First/Stop/Restart | Rebuild only the defined prefix; no future/previous-run cache leaks. Definitions survive same-archive navigation; Stop/First clear values to before-first, not retained future measurements. |
| C-04 | Equal-time records, all timestamps zero, empty archive and non-signal operations | Time seek/timed play apply complete due-time groups in sequence; manual Next applies exactly one. Empty/zero duration has no division error or fabricated sample; non-signal records change cursor/inspection, not signal values. |
| C-05 | 0.25x/1x/1000x, controlled waits and mid-interval pause/resume | Pacing scales captured elapsed intervals only; pause freezes the exact playhead and resume waits only the remaining interval. Replayed data/results are rate-independent. |
| C-06 | Repeated Play, pending Pause/Stop cleanup, close/unmount, delayed old wake or scheduler fault | One owner; repeated active Play is idempotent; competing start/navigation rejects before mutation during drain; no stale publication. Fault is visible, cleanup still works and no orphan Playing state remains. |
| C-07 | Invalid/unloaded/closed command, rate NaN/infinity/outside range, invalid seek/ordinal/definition | Typed visible rejection before mutation; previous data/context/definitions/position/rate/ownership intact. No clamping a bad request into success. |
| C-08 | Incomplete/Error source capture or end of its retained data | Prominent original completion/reason remains; explicit per-load acknowledgement required before Play/Next/seek. Reaching playback end never turns the source capture Complete or invents lost samples. |
| C-09 | Positive/failed read.signals, Unknown/Uncertain/non-numeric/missing/unit-change states | Shared accepted source/quality/range/last-Good rules apply; last-known data is clearly recorded Stale with original sample time. Only the affected module changes; no future fill, interpolation, false zero or fabricated reply. |
| C-10 | Zero-to-eight numeric/bar gauges and global unit changes | Reuse shared validated configuration/projection/rendering and accepted conversion/draft rules; exact source units/bounds remain canonical, invalid edits atomic, hides/remounts do not own playback |
| C-11 | Failed/cancelled/stale replacement load or overlapping command | Previous validated replay remains intact; replacement allowed only while quiescent. Old load/wake cannot publish after close or a newer generation. No failed load is disguised as a healthy empty replay. |
| C-12 | Current Replay versus original archived Simulation/Physical metadata | All primary controls/readouts are clearly recorded, never live/current vehicle evidence; no diagnostic factories/controls, catalogue backfill, model rerun or recording-workspace rebind |
| C-13 | Replay operations and live coexistence | Archive bytes/header/values/units/quality/times/raw payloads stay identical; store Save and live diagnostic/physics calls remain zero. Existing 313 regressions and nine-project Release build remain intact. |
| C-14 | Actual Client route and saved static-v1/dynamic-v2 captures | Real storage/codec/Application/UI path supports load/play/pause/navigation/units/hide/end/close, including restart without original profiles; native observations distinguished from controlled timing/fault tests |
### A. Archive-only context and load lifetime
Add a small factory/feature-owned replay session and a `/telemetry-replay` route
with shared navigation. The page owns replay lifetime; closing/unmounting it
cancels/drains playback and pending load. Individual gauge components do not.
Re-entering the page can create a fresh replay session; no disposed scoped
singleton must be reused accidentally.
List/load by explicit saved capture ID through the existing real archive store
and strict reader. Do not use the recording workspace's current live/unsaved
capture as a substitute or change its binding. Loading never auto-plays.
Ordinary live-host cleanup on route navigation stays as previously intended;
replay actions themselves never operate a live source.
A successful new load resets cursor/data/definitions/partial acknowledgement
and establishes a new replay identity. Keep valid runtime speed and global
display preference. Failed validation/I/O preserves the old replay and exposes
the actual error. Loading/replacing while Playing or draining is explicitly
rejected: pause/stop and finish cleanup first. Busy load rejects competing
commands; cancellation/disposal invalidates its generation before late results.
Loaded source completion must be visible throughout. For Incomplete or Error,
require a clearly labeled explicit acknowledgement bound to that load/capture
before data-producing playback/navigation; inspection/Stop/Close remain allowed.
Do not rewrite completion, original target or origin to label replay.
### B. Deterministic recorded prefix and navigation
Core playhead is integer TimeSpan ticks in [0, duration], where duration is the
last observation's captureElapsedTicks, or zero when empty. UTC start/stop,
diagnostic ticks and model time are original metadata, not playback duration.
Do not invent an unrecorded trailing interval from UTC or model time.
Expose applied observation count (0..N) separately from time. The loaded/rewound/
stopped cursor is **before first**: count 0, position 0, no published values.
This differs intentionally from SeekTime(0) if zero-time observations exist.
- Next applies exactly the next observation; Previous reconstructs one fewer
applied observation. Ordinal navigation reconstructs exactly the requested
count, including zero. Set playhead to that last record's elapsed time (or 0).
- Time seek applies every observation with elapsed time <= the requested tick,
including all ties, and preserves the exact requested playhead between records.
- Time-driven play applies all due observations in archive sequence before
publishing a coherent prefix. Equal-time records never reorder by module,
diagnostic ticks or model time.
- Manual navigation requires a fully quiescent Stopped/Paused/Completed player
and leaves Paused unless the entire nonempty archive is applied, then Completed.
- Stop cancels/drains, then resets to before-first while retaining definitions,
loaded archive, speed and acknowledgement. First/Rewind performs the same data
reset while already quiescent. Restart is rewind-plus-Play, also quiescent only.
Play resumes Stopped/Paused; Completed requires explicit Restart or navigation.
- Empty loaded archive has duration zero and no samples; Play/Restart completes
immediately without a timer. Seek(0)/Stop/First remain an empty stopped cursor;
Next/Previous has a visible no-records rejection. With all records at time zero,
Play/Seek(0) applies all; manual Next can pause inside that group. Completion
depends on applied count, not position==duration alone.
Use prefix reconstruction/reduction bounded by the accepted maximum 4,096
observations. Backward seek resets current/last-Good/module-status caches before
replaying the prefix. Never keep a later module value when earlier time has not
supplied it.
Pre-index binding metadata from actual positive snapshots in the loaded archive,
using the same first-observed module/signal/unit/origin rules as accepted live
data. This permits configuration before playback and keeps valid definitions
editable across seeks. The immutable metadata index contains no current values
and does not authorize publishing a future sample. Keep definitions across
navigation but show no-sample state when the applied prefix has not supplied
their data. Never infer values/bindings from the current vehicle catalogue.
Decisive fixture (10,000,000 ticks = 1 second):
| Sequence | Capture time | Actual archived operation |
|---|---|---|
| 1 | 1 s | module A read.signals: 850 rpm, Good |
| 2 | 2 s | module B read.signals: 12.6 V, Good |
| 3 | 2 s | module A read.signals: 1000 rpm, Uncertain |
| 4 | 4 s | module A read.signals: Timeout, no decoded successful signals |
Seek 1.5 s applies only #1: A=850 Good, B has no sample. Seek 2 s applies #1..#3:
A=1000 Uncertain, B=12.6 Good. Seek 4 s applies all: A retains **850**, the last
Good sample, visibly Stale with #1's time/reason; B remains 12.6 Good.
Previous from #4 reconstructs #1..#3 and restores Uncertain 1000, not Stale.
Before-first clears both. No interpolation/extrapolation/smoothing is assigned.
Reduce failures only for the affected read.signals module; probe/identity/DTC/
clear observations advance inspection/time but do not manufacture signal data.
Keep actual source quality/time/config/run and raw outcome. Unit/origin changes,
missing/non-numeric/Unknown values and no-prior-Good errors follow accepted
dashboard semantics. Archive facts themselves are never mutated.
### C. One cancellable playback owner
Speed defaults to 1x, accepts finite 0.25x..1000x inclusively and is runtime-only.
Change it only while quiescent; while Playing/draining, reject with visible
"pause first" behavior and unchanged rate/control value. Use proper binding
feedback for rejected UI changes, not a second uncommitted settings state.
Use monotonic TimeProvider plus a small cancellable wait seam for deterministic
tests; no new clock project/package. Elapsed wall time * selected speed advances
the capture-time playhead. Publish states consistent with the applied prefix;
smooth animation and a hard real-time SLA are not required.
Pause freezes playhead at the request instant, applies the due prefix and
cancels/drains the old wait without holding a gate needed by cleanup. Resume
rebases from that position, not from the last sample timestamp.
For example, from 1.5 s at 2x, the 2 s group is due after 0.25 wall seconds.
Pausing after 0.10 wall seconds freezes at 1.7 s; resume needs 0.15 more seconds.
Repeated Play while Playing shares the active owner. Any new owner/navigation/
replacement eligibility check precedes state mutation, including after end of
data while prior owner cleanup is pending. Use generation/owner guards on every
late wait/load publication; no orphan Playing state or duplicate frame.
Preserve one shared cleanup completion for repeated Stop/Close as appropriate.
Invalid inputs leave prior state intact except explicit error notification.
Clock/wait failures become visible Faulted replay with the last committed prefix
and original sample state intact, not successful end-of-data. Stop/Close still
drains/releases resources; recovery is explicit after cleanup.
Use overflow-safe tick arithmetic and bounded individual waits for large gaps,
with responsive cancellation, not a negative wrap or silent tail truncation.
### D. Shared presentation and actual UI
Reuse/extract small pure gauge definition-validation/projection/rendering helpers
from the existing live feature; do not duplicate them or create a dummy live
feed/session. Preserve live behavior through its existing tests. The replay
context is distinct from live DiagnosticSessionIdentity even if shared value/
definition models are reused.
Provide explicit archive selection, partial acknowledgement, Play/Pause/Stop/
Restart/First/Next/Previous, elapsed-time and ordinal seek, speed, playhead/count/
duration and source completion. Numeric/bar gauges retain the accepted in-memory
0..8 editor, exact unit binding, canonical bounds, precision and pinned draft
rules. Global units project display only; no archive/layout/cursor persistence.
All gauges say recorded/Replay, including Good quality, paused/completed state
and last-known values. Show original capture/provider/profile/session/target/
config/run, observation ordinal/capture time and per-sample diagnostic/model
time without implying simultaneous modules or a current live vehicle.
Raw/decoded inspection shows the current applied observation, with explicit
before-first/empty state. Navigation changes the replay prefix, not live state.
Reaching the end of partial data means "end of retained data", not source Complete.
Hiding/remounting gauges or unit changes cannot start/stop/seek/read the player.
Only the replay feature owns its clock/lifetime. Keep keyboard/accessibility/
safe text/error presentation and the existing routes/controls intact.
## FILES OR PROJECTS EXPECTED TO CHANGE
Application replay/context/controller/reduction, minimal pure shared gauge helper
extraction, UI replay route/navigation and shared renderer/editor adaptation,
composition only as needed, focused tests and related README/docs.
No new project/dependency, Simulation/Core physics/oracle implementation change,
diagnostic transport/provider or archive schema change.
Under `prompts`, create only `Step-08-Report.md`.
## INTERFACE AND DEPENDENCY REQUIREMENTS
Keep all nine project references and one Client, generic Components, archive-v1
strictness/limits, source bytes/types/units/quality/time, unit preferences, source
capture lifetime and existing diagnostic/partial-discovery/verified-clear gates.
Current kind is RecordedReplay; original source kind remains immutable metadata.
No source/hardware permission, catalogue backfill, arbitrary raw-command execution,
cloud/AI, broad catches/silent fallback, invented samples or success-shaped errors.
No package/system/workload install/update/repair/rollback, staging/commit/push
or destructive Git. Preserve protected/history/orchestration files and dirty work.
## TESTING REQUIREMENTS
Map C-01..C-14 to named meaningful tests before handoff:
- Build real validated saved v1/v2 archives through the accepted recorder/codec/
temporary filesystem. Then close the sources; replay without original profiles/
configurations. Use a spy/failing live-port composition and archive Save counter
to prove zero live operations/model calls/file rewrites.
- Assert the four-record fixture, between-time/backward/ordinal seeks, ties,
all-zero duration, empty/non-signal data and cleared future value/status caches
exactly. Verify metadata-only pre-indexing permits configuration before playback
without publishing any future measurements.
Test recorded reset/repeated model time without sorting or reconstructing it.
- Controlled monotonic time/waits prove rates, mid-gap pause/resume, delayed old
wakes, held cleanup, duplicate Play, generation replacement/unmount and faults.
Barriers establish ordering; no sleeps/polling or elapsed-performance proxies.
Release resources in finally and do not hold cleanup's gate across awaits.
- Real DI/rendered controls/editor verify invalid requests/rejections are visible
and atomic, partial acknowledgement/end labels, shared units/range/Unknown/
Uncertain/Stale/source context and no data/control action on gauge rerender.
- Compare archive bytes and canonical source metadata/definitions before/after
repeated playback/seek/unit operations. Preserve the 313 baseline cases;
map necessary UI-only assertion changes without deleting safety/error checks.
Run focused cases, then sequentially:
- `dotnet build .\src\AutoForge.slnx --configuration Release --no-restore`
- `dotnet test .\src\AutoForge.slnx --configuration Release --no-build`
Build all nine projects/native Client. Install/restore nothing absent a genuine
chosen-command missing dependency.
Native smoke: real saved static/dynamic captures, no live source open, new route
load/step/play/pause/seek/backward/units/gauges/speed/end/partial warning and route
close/re-enter. Restart the app and replay again without reimporting profiles.
Normal native timing is not controlled interleaving/fault proof. Verify the real
app storage root; do not assume LOCALAPPDATA redirects MAUI app data. Clean only
exact task-owned captures/profiles; preserve unrelated files/preferences.
## DOCUMENTATION REQUIREMENTS
Update README/index/architecture/build guide and replay/dashboard usage: explicit
RecordedReplay provenance, before-first versus zero-time seek, tie/prefix/last-Good
rules, time sources, rate/owner/drain/replacement policy, partial acknowledgement,
source immutability and actual native/controlled evidence.
Step-07/07A/07B is accepted; Step-08 worker status stays pending orchestrator review.
No interpolation, new recording durability, calibration/AI/hardware or physical
vehicle verification is claimed.
## ACCEPTANCE CRITERIA
| ID | Criterion | Evidence |
|---|---|---|
| AC-01 | C-01/C-11/C-12 explicit strict offline load and RecordedReplay provenance, immutable source and atomic replacement | Real file/store/context tests with no catalogue/live fallback and meaningful failed-load assertions |
| AC-02 | C-02..C-04/C-08/C-09 exact prefix/navigation/ties/empty/partial/quality behavior | Independent fixture expected values/counts/times/states, no future fill or fabricated samples |
| AC-03 | C-05..C-07 single cancellable owner, correct rate/pause/resume/drain/fault and rejection behavior | Controlled time/barrier lifecycle tests with exact playhead and prefix assertions |
| AC-04 | C-10 shared gauges/editor/units and honest original/current context in the real UI | Real rendered command/editor failures, preserved canonical bounds/drafts/source fields |
| AC-05 | C-13/C-14 all 313 regressions, meaningful new coverage, complete nine-project Release build/native playback pass | Exact commands/exits/warnings/per-project counts, zero-live/write spies and native/control distinction |
| AC-06 | Frozen matrix/self-review/docs/README describe actual output/state/time semantics and limits | Named executed cases and own post-green source-path review across producers/consumers |
| AC-07 | History/protection/dependencies/manual workflow preserved; report saved and worker stops | Preservation review and `prompts/Step-08-Report.md` |
## OUT-OF-SCOPE WORK
Resimulation, live diagnosis/recording changes, hardware/programming/calibration/
AI, interpolation/extrapolation/charts, layout/config/cursor persistence,
archive migration/edit/export or new schema/limits/streaming durability,
new project/package/host and unrelated cleanup.
No future assignment, agents, automatic handoff, orchestration/shared-gate edit,
stage, commit or push.
## REPORTING REQUIREMENTS
Save `prompts/Step-08-Report.md` wholly inside exactly one plain triple-backtick
fence, no language tag/outside text/nested fences. Include identifier/title/status,
functions/files/interfaces, C-01..C-14/AC-01..AC-07 matrix with named executed
evidence, exact prefix/playhead/rate/negative/lifecycle outcomes and zero-live/
write proof, commands/exits/warnings/per-project counts, native versus controlled
coverage, docs/README, own post-green source-path self-review, limitations/
deviations/blockers and preservation review.
COMPLETE requires all assigned checks; otherwise report INCOMPLETE/BLOCKED with
the precise unmet criterion. Save and STOP; do not begin Step-09.

View File

@ -179,6 +179,8 @@ public sealed class DiagnosticSession(
public ValueTask DisposeAsync()
{
IAsyncDisposable[] resources;
TaskCompletionSource disposalCompletion;
Task disposalTask;
lock (_lifetimeGate)
{
if (_disposeTask is not null)
@ -189,8 +191,26 @@ public sealed class DiagnosticSession(
Interlocked.Exchange(ref _closed, 1);
resources = _lifetimeResources.ToArray();
_lifetimeResources.Clear();
_disposeTask = DisposeResourcesAsync(resources);
return new ValueTask(_disposeTask);
disposalCompletion = new TaskCompletionSource(TaskCreationOptions.RunContinuationsAsynchronously);
disposalTask = _disposeTask = disposalCompletion.Task;
}
_ = CompleteDisposalAsync(resources, disposalCompletion);
return new ValueTask(disposalTask);
}
private async Task CompleteDisposalAsync(
IReadOnlyList<IAsyncDisposable> resources,
TaskCompletionSource disposalCompletion)
{
try
{
await DisposeResourcesAsync(resources).ConfigureAwait(false);
disposalCompletion.TrySetResult();
}
catch (Exception exception)
{
disposalCompletion.TrySetException(exception);
}
}

View File

@ -783,6 +783,130 @@ public sealed class TelemetryRecordingIntegrationTests
Assert.Equal(0, GetLifetimeResourceCount(session));
}
[Theory]
[InlineData(false)]
[InlineData(true)]
public async Task SessionCloseDuringCaptureDetach_DoesNotDeadlockOrLoseFailurePrefix(bool reset)
{
using var directory = new TemporaryDirectory();
var clock = new ManualTimeProvider();
using var provider = CreateProvider(directory.Path, clock);
var (session, protocol) = await OpenBarrierSessionAsync(provider);
var limits = reset
? TelemetryCaptureLimits.Default
: TelemetryCaptureLimits.Default with { MaximumObservations = 1 };
var workspace = new TelemetryRecordingWorkspace(
provider.GetRequiredService<ITelemetryArchiveStore>(),
clock,
limits);
var barrier = new SubscriptionDisposeBarrier();
protocol.ArmNextSubscription(barrier);
workspace.Start(session);
var endpoint = EngineEndpoint(session);
Assert.True((await session.ReadSignalsAsync(endpoint)).IsSuccess);
if (reset)
{
clock.Advance(TimeSpan.FromTicks(-1));
}
var operation = Task.Run(() => session.ReadSignalsAsync(endpoint));
Task? closeOrReset = null;
DiagnosticSession? replacement = null;
try
{
await barrier.CallbackDisposeEntered.Task.WaitAsync(TimeSpan.FromSeconds(5));
closeOrReset = reset
? Task.Run(async () =>
replacement = await provider.GetRequiredService<IDiagnosticSessionService>()
.ResetAsync(session))
: Task.Run(async () => await session.DisposeAsync());
await barrier.ClosingDisposeEntered.Task.WaitAsync(TimeSpan.FromSeconds(5));
}
finally
{
barrier.ReleaseCallbackDispose.TrySetResult();
}
try
{
var dependentWork = Task.WhenAll(operation, closeOrReset!);
await Task.WhenAny(dependentWork, Task.Delay(TimeSpan.FromSeconds(5)));
Assert.True(
dependentWork.IsCompleted,
$"Operation completed={operation.IsCompleted}; close/reset completed={closeOrReset!.IsCompleted}.");
Assert.True((await operation).IsSuccess);
await closeOrReset!;
var snapshot = Assert.IsType<TelemetryCaptureSnapshot>(workspace.CurrentCapture);
Assert.Single(snapshot.Observations);
Assert.Equal(1, snapshot.Completion!.ObservationCount);
Assert.Equal(
reset ? TelemetryCaptureState.Error : TelemetryCaptureState.Incomplete,
snapshot.Completion.State);
Assert.Equal(
reset
? TelemetryCaptureStopReason.ObservationFailure
: TelemetryCaptureStopReason.LimitExceeded,
snapshot.Completion.StopReason);
Assert.Equal(0, GetLifetimeResourceCount(session));
if (reset)
{
Assert.NotNull(replacement);
Assert.NotEqual(session.Identity.SessionIdentifier, replacement.Identity.SessionIdentifier);
Assert.True((await replacement.ReadSignalsAsync(EngineEndpoint(replacement))).IsSuccess);
Assert.Single(snapshot.Observations);
}
else
{
Assert.True(session.IsClosed);
}
}
finally
{
barrier.ReleaseCallbackDispose.TrySetResult();
if (replacement is not null)
{
await replacement.DisposeAsync();
}
}
}
[Fact]
public async Task ConcurrentSessionDispose_SharesCompletionAndReportsCleanupFailure()
{
using var directory = new TemporaryDirectory();
using var provider = CreateProvider(directory.Path);
var session = await OpenStaticSessionAsync(provider);
var resource = new ControlledDisposalResource(
new InvalidOperationException("controlled cleanup failure"));
session.RegisterLifetimeResource(resource);
var firstDispose = session.DisposeAsync().AsTask();
Task? secondDispose = null;
try
{
await resource.DisposeEntered.Task.WaitAsync(TimeSpan.FromSeconds(5));
secondDispose = session.DisposeAsync().AsTask();
Assert.Same(firstDispose, secondDispose);
Assert.Throws<ObjectDisposedException>(() =>
session.RegisterLifetimeResource(new ControlledDisposalResource(null)));
Assert.Equal(1, resource.DisposeCallCount);
}
finally
{
resource.ReleaseDispose();
}
Assert.NotNull(secondDispose);
var firstFailure = await Assert.ThrowsAsync<InvalidOperationException>(() => firstDispose);
var secondFailure = await Assert.ThrowsAsync<InvalidOperationException>(() => secondDispose!);
Assert.Same(firstFailure, secondFailure);
Assert.Equal("controlled cleanup failure", firstFailure.Message);
Assert.Equal(1, resource.DisposeCallCount);
}
[Fact]
public async Task CaptureCountLimit_Accepts4096OperationsAndMarksTheNextOperationIncomplete()
{
@ -1320,6 +1444,35 @@ public sealed class TelemetryRecordingIntegrationTests
ExecutionTargetKind.Simulation);
}
private static async Task<(DiagnosticSession Session, BarrierObservationProtocolClient Protocol)>
OpenBarrierSessionAsync(IServiceProvider provider)
{
var prototype = await OpenStaticSessionAsync(provider);
var profile = prototype.Profile;
var identity = prototype.Identity;
await prototype.DisposeAsync();
var factory = provider.GetServices<IDiagnosticTransportFactory>()
.Single(candidate => StringComparer.OrdinalIgnoreCase.Equals(
candidate.Descriptor.Target.ProviderIdentifier,
identity.Target.ProviderIdentifier));
var transport = factory.OpenTransport(profile, identity);
var protocol = new SyntheticDiagnosticProtocolClient(
profile,
identity,
transport,
new DiagnosticLogicalClock());
var barrierProtocol = new BarrierObservationProtocolClient(protocol);
return (
new DiagnosticSession(
profile,
identity,
barrierProtocol,
transport,
new CancellationTokenSource()),
barrierProtocol);
}
private static int GetLifetimeResourceCount(DiagnosticSession session)
{
var resources = (ICollection)typeof(DiagnosticSession)
@ -1708,6 +1861,111 @@ public sealed class TelemetryRecordingIntegrationTests
_inner.LoadAsync(captureId, cancellationToken);
}
private sealed class SubscriptionDisposeBarrier
{
public TaskCompletionSource CallbackDisposeEntered { get; } =
new(TaskCreationOptions.RunContinuationsAsynchronously);
public TaskCompletionSource ClosingDisposeEntered { get; } =
new(TaskCreationOptions.RunContinuationsAsynchronously);
public TaskCompletionSource ReleaseCallbackDispose { get; } =
new(TaskCreationOptions.RunContinuationsAsynchronously);
}
private sealed class BarrierObservationProtocolClient(
SyntheticDiagnosticProtocolClient inner) :
IDiagnosticProtocolClient,
IDiagnosticOperationObservationSource
{
private readonly IDiagnosticOperationObservationSource _observationSource = inner;
private SubscriptionDisposeBarrier? _nextSubscriptionBarrier;
public DiagnosticSessionIdentity Session => inner.Session;
public IReadOnlyList<DiagnosticEndpoint> ExpectedEndpoints => inner.ExpectedEndpoints;
public DiagnosticTraceSnapshot Trace => inner.Trace;
public int ProtocolVersion => _observationSource.ProtocolVersion;
public string? ObservationDispatchError => _observationSource.ObservationDispatchError;
public void ArmNextSubscription(SubscriptionDisposeBarrier barrier) =>
Interlocked.Exchange(ref _nextSubscriptionBarrier, barrier);
public IDisposable Subscribe(Action<DiagnosticOperationObservation> observer)
{
var subscription = _observationSource.Subscribe(observer);
var barrier = Interlocked.Exchange(ref _nextSubscriptionBarrier, null);
return barrier is null
? subscription
: new BarrierSubscription(subscription, barrier);
}
public Task<DiscoveryResult> DiscoverAsync(CancellationToken cancellationToken = default) =>
inner.DiscoverAsync(cancellationToken);
public Task<DiagnosticOperationResult<ModuleIdentityResult>> ReadIdentityAsync(
DiagnosticEndpoint endpoint,
CancellationToken cancellationToken = default) =>
inner.ReadIdentityAsync(endpoint, cancellationToken);
public Task<DiagnosticOperationResult<TroubleCodeReadResult>> ReadTroubleCodesAsync(
DiagnosticEndpoint endpoint,
CancellationToken cancellationToken = default) =>
inner.ReadTroubleCodesAsync(endpoint, cancellationToken);
public Task<DiagnosticOperationResult<TroubleCodeClearResult>> ClearTroubleCodesAsync(
DiagnosticEndpoint endpoint,
CancellationToken cancellationToken = default) =>
inner.ClearTroubleCodesAsync(endpoint, cancellationToken);
public Task<DiagnosticOperationResult<SignalSnapshot>> ReadSignalsAsync(
DiagnosticEndpoint endpoint,
CancellationToken cancellationToken = default) =>
inner.ReadSignalsAsync(endpoint, cancellationToken);
private sealed class BarrierSubscription(
IDisposable innerSubscription,
SubscriptionDisposeBarrier barrier) : IDisposable
{
private int _disposeCalls;
public void Dispose()
{
if (Interlocked.Increment(ref _disposeCalls) == 1)
{
barrier.CallbackDisposeEntered.TrySetResult();
barrier.ReleaseCallbackDispose.Task.GetAwaiter().GetResult();
}
else
{
barrier.ClosingDisposeEntered.TrySetResult();
}
innerSubscription.Dispose();
}
}
}
private sealed class ControlledDisposalResource(Exception? failure) : IAsyncDisposable
{
private readonly TaskCompletionSource _releaseDispose =
new(TaskCreationOptions.RunContinuationsAsynchronously);
private int _disposeCallCount;
public TaskCompletionSource DisposeEntered { get; } =
new(TaskCreationOptions.RunContinuationsAsynchronously);
public int DisposeCallCount => Volatile.Read(ref _disposeCallCount);
public void ReleaseDispose() => _releaseDispose.TrySetResult();
public async ValueTask DisposeAsync()
{
Interlocked.Increment(ref _disposeCallCount);
DisposeEntered.TrySetResult();
await _releaseDispose.Task;
if (failure is not null)
{
throw failure;
}
}
}
private sealed class TestNavigationManager : NavigationManager
{
public TestNavigationManager(string uri)