jobtrackingapp/docs/redesign/phase-9/p9-05/README.md

# P9-05 — Documents, edge cases and finite performance

The final separate Phase 9 diagnostic lane completed **18/18**, zero failures/skips/flaky/retries (`phase9-cta`, exit 0 at 03:18:12 UTC on 23 October). It includes 12 route cases, four accessibility cases, one RD21 characterization and one bounded workload. **A passing characterization does not qualify the unsafe behavior it records.** Earlier reports remain retained as superseded evidence.

The earlier diagnostic snapshot is retained as [source-manifest-before-admin-fixture.json](../p9-00/source-manifest-before-admin-fixture.json). The [Admin locator delta](../p9-00/admin-fixture-delta.json) and [provider-worker fixture delta](../p9-00/worker-isolation-delta.json) retain the two subsequent fixture changes. Product/export/core inputs are unchanged; inherited diagnostics rerun after the worker configuration repair.

## Native documents

The final oracle invocation is bound to the completed first ordinary shard by [native-document-provenance.json](evidence/native-document-provenance.json): 93 input hashes include its report/exit, the unchanged oracle, measurements, exported PDFs and preview images. Both oracle commands exited 0; inputs remained stable across execution. Portable result hashes are verified again by the reporting script.

Fresh real preview and native exported PDF bytes from the retained corpus passed **16/16 fixtures /36 pages** under the unchanged oracle: exact page count/membership, physical dimensions ±0.5mm, marker positions ±2mm, font sizes ±0.5pt and original RGB tolerance. Nine defaults, long EN/NB, 8mm Letter, 28mm roomy, grid/bubbles, long sidebar and unbreakable boundary fixtures are included. No renderer geometry or tolerance changed.

[Portable oracle results](evidence/fidelity-results.json) retain all measurements. The final isolated ordinary shard's native corpus producer passed; the unchanged oracle passed 16/16 over its fresh 36 pages. All 36 regenerated page thumbnails were inspected in [sheet1](evidence/pdf-sheet-1.png), [sheet2](evidence/pdf-sheet-2.png) and [sheet3](evidence/pdf-sheet-3.png), plus the full-size preview/PDF `unbreakable-and-boundary/comparison-4.png`. The final regenerated sheets are byte-identical to the inspected sheets. No visible clipping/overlap/blank output was found in those views. Several last pages contain little content; that is layout behavior, not missing markers. Thumbnail/spot inspection does not establish print/readability, every photo/glyph or intended Linux-font acceptance. All 362 ordinary cases completed separately.

Existing browser fixtures also cover hidden preview mounting, image-layout readiness, dirty edits, stale CV revisions, preview transport recovery, canonical versus variant overrides, frozen submitted packages/archived binaries and missing optional sections. Their full-suite outcomes belong to final regression. Native PDF admission uses inert child processes and is not added to fidelity counts.

The separate Settings corpus passed the same unchanged oracle over **8 fixtures /24 pages**: EN/NB × light/dark, before and after actual document-control changes. All24 page thumbnails were inspected in [Settings sheet1](evidence/settings-pdf-sheet-1.png) and [sheet2](evidence/settings-pdf-sheet-2.png); a full-size light-EN before preview/PDF comparison was also inspected. No visible clipping/overlap/blank output was found in those views. [Measured results](evidence/settings-fidelity-results.json) retain the before/after page dimensions, marker/font/colour agreement. These are separate24 pages, not extra fixtures in the16/36 baseline corpus, and do not close human/Linux/print requirements.

## Sustained local workload

[Raw finite measurements](performance.json): **63.332 s**, two paced readers, **240 successful reads**, 250 owned applications, eight Career roles with 48 distinct long bullets, and 20 native navigations across Dashboard/Table/Board/long CV. The saved canonical profile and exact 250-job membership remained intact afterward. CV readiness waits for settled pagination and verifies the final bullet's page membership and multiple rendered pages before recording timing.

API latency: **p50 12 ms /p95 42 ms /max 48 ms**. Native ready ranges: Dashboard 942–977 ms; Table 981–1035 ms; Board 951–966 ms; long CV 1492–1549 ms. These include browser boot/auth/render work and automation overhead. The host reports Windows, 16 logical CPUs, 34,145,554,432 bytes RAM and Chromium 151.0.7922.34; API uses isolated SQLite and frontend uses the rebuilt static export. Jest had finished before this final run. It remains a shared development host, not an isolated machine comparison. Earlier 65.303 s /p95 76 ms and 62.890 s /p95 39 ms checkpoints remain in superseded raw reports and are not combined with these measurements.

CDP CPU/task/heap/DOM metrics are retained for every navigation. They are observations, not leak diagnosis or attributed backend resource profiling. Two paced readers for a minute do not establish capacity/SLA, sustained drag frames, physical mobile performance, long-term memory stability, MariaDB/Linux behavior or real model inference throughput. No production load was generated.

Retained ordinary fixtures extend edge-case coverage to no-data,1,000-job Board membership/filtering, long timelines/profiles, revision conflicts, stale sessions, HTTP/network failures, missing/foreign references and recoverable drafts. Remaining worker durability, creation idempotency and high RD21 stale writes are explicit risks. No synthetic data was loaded into production.

A separately isolated [volume supplement](volume/README.md) adds3 passing native API cases:100 CV revisions plus additive restore101,100 long manual timeline messages with exact membership and foreign-owner denial, and100 cancellation notices with bounded listing/read/dismissal counts. Its source hashes and results are separate from the825-input ordinary-run snapshot. This establishes finite persisted membership/count/revision behavior, not large-history browser capacity.