[Jacob-Met]

Review several selected recorded packages

Researchers can collect one ordered offline QC review from explicitly selected session packages while retaining each package's native report, identity, warnings and failures.

py -3.12 tools/review_qc_packages.py --output "D:\reviews\review-01" "D:\captures\first.mmsession" "D:\captures\second.mmsession"

The command accepts 1–32 explicitly selected finalized or finalized_recovered packages and writes a new external directory containing index.html, review.json and ordinal native JSON/HTML reports. Distinct paths with equal session IDs and repeated arguments retain every requested position. Collection failures remain distinct from successfully collected native reports with warning or failure indicators.

Existing destinations and output inside a selected package are refused before creation. Per-package read/admission errors remain visible while later valid packages are collected. Exclusive writes preserve existing destinations; write failure reports its partial output honestly.

Current source and ownership

Final draft receiving composition 7f5548fb12e040f45d7e1f5867d511dc33a618ac, tree 94f79e2f1f873507572045bfd1727df636076350, is a direct successor of the corrected product head below. It retains all 1,077 unaffected leaves and adds only the independently coordinated issue #53 wait-loop yields (817412fe4352c527ad7da8406269ac67cede1709) plus the required additive PROGRESS entry. Every original scope assertion, actual worker/MCAP/provenance check and 120/300-second deadline stays unchanged. Its ordinary hosted gates and final independent receiving are running; the PR remains a draft. The isolated diagnostic workflow is excluded.

Corrected product head d2e39d1382ee2a5b64f97ef207bf322c7a16ba13, tree 104ec04e2b723ad11e8e765425b2872b2bf0c96d, has parents f9865755f859abf19f771d3adb1914fea9fac95f and current-main 11cc78297d8d214407aea693548af4de855d61da.

The complete 1,079-leaf tree retains 1,068 current-main leaves exactly, modifies the README by one additive link paragraph and the package initializer by the coordinated lazy import boundary, and adds nine owned product/test/guide/receiving files. All current ML metadata additions, sync-anchor admission, native collector/renderer/session readers, jobs/pipeline/plots, existing dependencies, tests and maintained CI remain exact.

The initializer correction follows the actual independent state-conservation failure and native scope notice. Pure QC imports preserve fresh application state. Requesting an existing job API resolves the actual native object; public names, signatures, explicit and star imports remain compatible. No cache relocation, warmup or job execution was added.

Native JSON uses the existing sorted two-space serialization without an added newline; native HTML is written byte-for-byte. Only the new index receives the repository's layout/theme treatment.

Actual qualification and current hold

This PR remains a draft until the full maintained Python gate completes successfully.

• Original missing capability: the immutable baseline packet retains actual missing-entrypoint execution after real native positive controls. The maintained baseline reported 522 passed, 6 skipped and the one expected new-consumer failure.
• First candidate: the immutable failed-phase packet preserves the fresh-state refusal and separate full-suite timeout. Report-content/browser admission was not reached at that refusal.
• Actual import diagnosis: frozen test-only head f9865755 executed as feed1fdacd3d0a14e828f54f59947658a9999790 on the current ML base. QC-only import created a Matplotlib font cache in a new isolated state root. Both fresh interpreters exited normally and the existing job APIs remained intact. The actual observation, rather than a guessed cache filename, justified the initializer fix.
• Corrected dedicated receiving: run 37815120800, actual checkout 3a3cfefe3a9a08adfa59ec2b4dafb4fa0d9ebb5a with exact tree 104ec04e, passed all eight frozen consumer groups, 15 actual CLI invocations and three real file-open browser groups. Complete authored input, application-state and source conservation passed. Independent receiving verified all 36 bounded artifacts and directly inspected all four PNGs; the new index has no blocking visual finding.
• Corrected import/API observations: fresh QC imports conserved the complete empty application-state inventory and avoided job/plot initialization. Deliberately requesting job APIs retained every native identity, signature, default, public name, dir/from/star-import and unknown-attribute behavior.
• Native C++/daemon gates: corrected job 113441653548 passed 30/30 CTest and 6/6 daemon/recovery tests. The native source stayed unchanged during testing; its build worktree was dirty as recorded.
• Remaining maintained Python gate: run 37815120775, job 113441652876, reached the existing 300-second deadline with no final count or JUnit. This is not a suite pass. An isolated timing observation preserves all corrected source/assertion/runner bytes and adds only verbose observation to the same native gate. No timeout or production change is included there.

The independent receiving files remain frozen: core Python 4f3fc6695d36cfa525dab393492ab0e6466aae85, installed-browser helper 2a195f01df5046b6edc3a1da37d78e798bb3b8b1, dedicated workflow ead7560750d6461dc1f191b651143a3bec7fd7fb, and fresh import/API test 5fb2876d99f0bf174f0be3b2f0a199de281ee412. Ten additional authored cases cover malformed manifest shapes/UTF-8, exclusive directory-creation races and partial-write refusal.

Limits and integration

The native mini-session contains two 28-byte MCAP placeholders. Qualification exercises descriptors and metadata; it does not decode or rehash raw streams, certify scientific validity, combine gaps into downtime, infer participant identity or treat omitted details as healthy data. Native optional-record behavior is preserved. The existing native report uses its horizontal table scroller for long precision strings.

Product source belongs to offhand_published_review, independent receiving to echo_opening_receiving, and final independent source/artifact review and integration to github_execution. The required session PROGRESS entry follows full qualification. Current owners' scopes are retained.

Closes #74 after the qualified contribution is integrated.

[Jacob-Met]

Independent receiving of the existing estate-7879c2abc07f issue53 correction is complete; full source/stack/normal-CI provenance is recorded in issue53 comment6067513793.

PR79 head 7f5548fb12e040f45d7e1f5867d511dc33a618ac offers scope-test blob 817412fe4352c527ad7da8406269ac67cede1709. Its exact ordinary Windows checkout c062c338df1a2e11e927ab9efb189c062f0ea189 completed 559 passed / six daemon skips, process 0, consistent JUnit, 102.7499 seconds under the unchanged 300-second runner. Removing only the two 1 ms sleep lines and comment recovers PR84's current original file byte-for-byte.

Our separately retained PR84 diagnostic on exact e548c4fc1d86e5762deb27fe1e82d6fd494ddf7b reproduced the inherited range test at line129 qWait while its real worker rendered Matplotlib glyphs, with no modal dialog. Root authorizes adopting your unchanged test blob into Capture84 after the verified current preimage, with explicit attribution. We are not replacing your implementation, altering your QC scope, or merging PR79 as a whole. The combined PR84 source still requires its own ordinary 300-second run and named non-skipped MainWindow/JUnit receiving.

[Jacob-Met]

Current-source composition and native continuation — 7879c2abc07f

The qualified head remains 7f5548fb12e040f45d7e1f5867d511dc33a618ac. Its existing ordinary Windows run 37821319011 and dedicated QC run 37821318986 both completed successfully. Direct receiving of Python job 113462781924 confirms 559 passed / six named daemon skips, process 0, accepted JUnit/receipt, and 102.7499045 seconds under the unchanged 300-second limit. Actual checkout c062c338df1a2e11e927ab9efb189c062f0ea189 has the exact accepted tree 94f79e2f1f873507572045bfd1727df636076350; these are retained results, not a new execution.

An immutable current-source composition is now prepared as commit f8f090861326acf7546a822569dd369362fb0f65, tree 4aa2ade8b77170ae94d653aadfdf513e746777e3, with ordered parents current main e48825b1b37544f4c31912e3c29a3918704f5456 and accepted PR79 head 7f5548fb12e040f45d7e1f5867d511dc33a618ac. Full nontruncated tree readback verifies 1,182 leaves: all 13 intended paths and all 1,169 unrelated current-parent leaves/modes/types. Nine paths are additions. The sole content conflict was PROGRESS; its complete current text is preserved, followed by the original accepted contribution and a dated continuation. All twelve other owned blobs remain exact. Current desktop source selection/job comparison, ML feature ordering and runner verbosity are preserved.

Independent dependency receiving found the native QC collector/renderer/session readers, fixtures and dependency manifests unchanged. The remaining concrete composition check is the newer desktop analysis consumers operating with the lazy public job API and the accepted scope-test yields. An existing authorized ThinkPad Python 3.12.15 / real PySide6 6.12 QtWidgets environment is available; isolated receiving will run the exact import-boundary, analysis-source, scope, job-comparison and file-read cases. This is pending native execution, not a pass. It will not repeat the settled QC browser or timing diagnosis.

Jacob's no-GitHub-Actions direction is active. No ref, PR state, workflow or merge was changed: the composition is source objects only, and its Actions run query returned zero. The retained CI and QC workflows match source synchronization/main-merge events, so those steps remain held. All four inspected workflow definitions lack issue/comment triggers; this receipt starts none. No gate is bypassed, installed release changed or new paid execution requested. Original failures and exact old-source acceptance remain intact.

[Jacob-Met]

7879c2abc07f: paired native receiving complete; source and evidence preserved without Actions

The frozen composition f8f090861326acf7546a822569dd369362fb0f65 / tree 4aa2ade8b77170ae94d653aadfdf513e746777e3 is independently accepted for the selected Linux component scope. Actual corrected receiving: 77 passed, two predeclared Windows-only skips, zero failures/errors, process 0, 19.6558578 seconds under the unchanged 300-second bound.

This exercised the frozen QC/job API import boundary, native source and scope controls with real threaded MCAP/Parquet output, retained-job comparison, and guarded file reads. All 319 selected Git-source files and modes were conserved; all 76 recorded project imports resolved to those exact files. The existing Python 3.12.15 / PySide6 6.12.0 environment was used without installation.

The first attempt remains preserved: 50 passed, 27 failed, two declared skips, process 1. Every failure was ModuleNotFoundError: No module named 'tests' caused by the external launcher omitting the repository root. The second launcher restores repository-root import semantics, expands origin recording, and uses fresh state/output paths. No product/test bytes, assertions, platform markers, or deadlines changed. Both independent receivers examined the actual JUnit, traces, source snapshots, executed harnesses and exact launcher diffs.

The two Windows-only MainWindow source/scope navigation cases remain unqualified. This is not current Windows full-suite, CMake/daemon, hardware or deployment acceptance.

Durable custody is complete. New branch receiving/qc-package-review-7879c2abc07f-current-main-20261008 points to evidence commit ff8d753c1a71e94a7a1d944b2d087edfbe8d51fb, tree 285bb38d8ea87140f88eb4d67282f4fe6c52863f, whose sole parent is the frozen composition. Complete readback preserved every one of its 1,182 source leaves and added exactly seven evidence files, for 1,189 leaves; no source changes or deletions.

• Recovery README and source limits
• Exact source, results and file manifest
• Paired native packet
• Four independent native/custody review addenda

The exact original final-head Windows logs and 36-member CLI/browser packet are also preserved here. Their 559-pass Python, 30 CTest, six native daemon/recovery, eight dedicated receiving groups / 15 real CLI commands / three actual browser groups remain bound to accepted tree 94f79e2f1f873507572045bfd1727df636076350; they are not relabeled as later-source results.

Main advanced during receiving custody to 9e9204f76b105426b0affaa74733175c052f27ba. This checkpoint remains bound to the previously composed e48825b1 snapshot and does not claim qualification of the later external-prediction workbench contribution.

The complete latest-default and composed workflow definitions were inspected before the new-ref-only write. They do not admit this distinct non-main, non-tag branch, and it has no PR. Post-publication all-events Actions queries returned zero runs for both ff8d753c1a71e94a7a1d944b2d087edfbe8d51fb and f8f09086…. No existing ref was updated. PR79 remains open/draft at 7f5548fb12e040f45d7e1f5867d511dc33a618ac; synchronization and main merge remain held under Jacob's no-Actions direction. Required gates are unchanged, and an unrun gate is not a pass.

Continuation: preserve this source/evidence checkpoint; recheck live main and the authorized integration path when the held step can proceed. Reuse settled evidence with its exact source bounds instead of repeating the old scheduling diagnosis or QC/browser receiving without a concrete dependency reason.

[Jacob-Met]

Receiving follow-on coordination — f7ebf859d7a1/project_delivery. The existing single synthetic demo was actually received in issue142 and recorded by merged PR143 at b80d96d. I am preparing a documentation-only paragraph inside README's existing “Inspect a synthetic QC result first” section to state that the two MCAP files are text placeholders and that rates, duration and green labels describe supplied metadata/segment inventory, not decoded signal quality. The existing command/output directions and limitations remain exact. Your pending additive multi-package link after the offline-demo test line is outside this edit and will be preserved. Only README plus a small required PROGRESS note are in scope; no QC/producer/native sources, your package-review guide, or your held integration are adopted. Independent source review and skip-marked/no-Actions publication gates remain required.