SPB Git

Repository Public License

RELEASE-REPORT.md 11.6 KB 171 lines markdown

1<!--2ORIX Monitor3File: RELEASE-REPORT.md4Author: Simon-Pierre Boucher5Contact: contact@spboucher.ai6-->78# ORIX Monitor 1.0.0 (build 1) — release report910Date: 2026-11-21 · Bundle identifier: `ai.spboucher.orix.monitor` · Minimum macOS: 15.0 · Architecture: arm641112**Summary.** The app is built, tested, signed with a Developer ID, notarized by Apple (app and disk image), stapled,13accepted by Gatekeeper, and was installed and run on two other Macs. The repository was kept private until the owner asked14for publication; on 2026-11-21 it was made **public** and release **v1.0.0** was published at15<https://git.spboucher.ai/orix-monitor/releases>. The published files were downloaded back and verified.1617## 1. Implemented features1819| Area | Delivered |20|---|---|21| Dashboard | 14 widgets, edit mode (add, remove, resize, reorder by menu or drag and drop), 3 built-in layouts, saved layouts, detached dashboard windows |22| CPU | Total / user / system / idle, per-core bars and core × time heatmap, load averages, live-window peak and average, core groups, top processes |23| GPU | Metal device capabilities; best-effort utilization, renderer, tiler, core count, driver memory |24| Memory | Used / app / wired / compressed / cached / free, pressure level and history, swap, paging rates, definitions, top processes |25| Disk | Volumes with tags and containers, block devices with throughput and totals, available-space trend |26| Network | Throughput, interfaces by kind, Wi-Fi link quality, DNS and routing, on-demand TCP latency test, change log |27| Processes | Sortable, filterable, searchable table; inspector with live history of the selected process; reveal, bring to front, copy PID, quit and force quit behind confirmation |28| Power | Charge, state, source, time estimates, Low Power Mode, adapter; best-effort health, capacities, power flow; power by process |29| Thermal | Thermal state and history; explicit "not available" rows; availability matrix probed on the running Mac |30| Menu bar | Configurable item (5 indicators, mini charts, alert dot) and quick panel |31| History | SQLite store with raw / minute / hour tiers, range and metric comparison, gap shading, bookmarks, PNG and CSV export, configurable retention, erase |32| Alerts | Rules with threshold, hold time, cooldown, severity; snooze; history; menu bar; optional Notification Center |33| Insights | 7 detectors over stored history, each finding with its evidence and a suggested action |34| Reports | 7 report types × PDF / Markdown / CSV / JSON; comparison of two JSON reports |35| Settings | Appearance, density, colour-blind-safe palette, launch at login, sampling intervals, battery saver, best-effort switch, retention, menu bar items, privacy mask |36| Localization | English and French (Canada), 747 strings, locale-aware formats |37| Icon | Five concepts explored, one chosen, full macOS icon set |3839## 2. Verified metrics4041Checked by the 81 core tests (arithmetic on fixed inputs, and invariants on the live machine) and by comparing42`orix-probe` with `top`, `vm_stat` and `ps` on the development Mac.4344| Metric | Verification |45|---|---|46| CPU total, user, system, per core | Tick arithmetic including 32-bit wraparound; fractions sum to 1; core count matches the kernel; totals consistent with `top` |47| Memory accounting | used + cached + free = physical on the live machine; saturating arithmetic on inconsistent counters; wired and compressed consistent with `top` |48| Memory pressure, swap | Level mapping; swap used ≤ swap total |49| Disk throughput, volumes | Rates from counters, reset handling; startup volume present with a BSD name; available ≤ total |50| Network throughput | 64-bit and 32-bit counter handling, resets give no rate; loopback present; interface classification |51| Processes | The test process finds itself with plausible CPU after burning CPU; CPU time agrees with `getrusage` within 5 %; `launchd` listed without statistics |52| Battery | Signed registry values, both registry layouts, derived watts and capacity ratio; plausible ranges on the live machine |53| GPU | Parsing and clamping of the registry statistics; devices listed by Metal |54| Thermal state | Level mapping |55| Store | Exact minute and hour aggregates, downsampling that preserves extremes and weighted mean, gap reporting, retention, relaunch merge, millisecond timestamps |56| Alerts, findings, reports | State machine, detectors, renderers |5758## 3. Unsupported metrics5960Declared in the app as "Not available through supported system interfaces." and never shown: CPU temperature, GPU61temperature, fan speed, CPU frequency, SSD health and wear, per-process network traffic, per-process GPU usage, whole-system62power on desktop Macs. Not requested: Wi-Fi network name (needs Location Services). Shown as identity only: processes of63other users. Battery temperature is absent on the development Mac under macOS 27 and is shown only where the gauge64publishes it; that display path has run on synthetic data only.6566## 4. Test results6768| Suite | Result |69|---|---|70| Core, `swift test` | **81 tests in 16 suites passed** |71| Interface, XCUITest | **13 executed: 12 passed, 1 skipped, 0 failed** |72| Accessibility audit | Ran on 12 pages; 440 notes recorded, none resolved (see §9 and `docs/reports/accessibility.md`) |73| Repository hygiene, `Scripts/check.sh` | Author headers present, no secret found |7475Skipped: `testMenuBarExtraOpensPanel` — macOS hides the status item behind the notch on the development Mac, so the item's76existence and live title were verified but it was not clicked.7778Defects found by testing and fixed before this build: (1) the app aborted on the Processes page in a narrow window; (2)79processes of other users were missing from the table; (3) the icon manifest was empty, so the app had no icon; (4) alert80hold timers were cleared at sampling intervals of about 30 s or more; (5) screenshot launch options could change real81preferences; (6) light-mode text colours were below 4.5:1; (7) chart animation cost 10 % of a core on the CPU page.8283## 5. Performance overhead8485Signed Release build on the development Mac, AC power, window visible, other heavy work running on the machine.8687| Page | Interval | CPU (share of one core) | Footprint at end of run |88|---|---|---|---|89| Dashboard | 5 s | 0.87 % | 269 MB |90| Dashboard | 2 s (default) | 2.32 % | 80 MB |91| CPU | 2 s (default) | 2.47 % | 240 MB |92| Processes | 2 s (default) | 2.14 % | 63 MB |93| CPU | 1 s | 4.13 % | 254 MB |9495Memory moves between about 60 MB and 270 MB within a run; about 185 MB of the high figure is graphics memory of the96window's layers, about 41 MB is heap. **The cause of that swing is not established.** Collector timings, store timings and97the full method are in `docs/development/performance.md`. Energy impact was not measured.9899## 6. Supported and tested configurations100101| Mac | macOS | What was done |102|---|---|---|103| MacBook Pro `Mac17,6`, M5 Max, 48 GB | 27.0 (26A5425a) | Development, all tests, benchmarks, screenshots, notarization |104| Mac mini `Mac14,3`, M2 | 26.5.1 | Final DMG: checksum, Gatekeeper with quarantine flag, 45 s run, stored metrics read back |105| Mac mini `Mac16,11`, M4 Pro | 26.5.1 | Same |106| Mac mini `Mac16,10`, M4 | 26.7 | Same, with an earlier build of 1.0.0 only |107108Declared minimum is macOS 15.0; **no Mac running macOS 15 was available**, so the minimum is the deployment target the app109compiles against, not a tested configuration. Intel is not supported (no Intel slice).110111## 7. Known limitations112113See README §32. In short: one fully tested Mac; no temperatures, fans or SSD health by design; identity only for other114users' processes; best-effort values may vanish; history has holes when the app is not running and can lose up to 30 s on115a crash; the menu bar item can be hidden by macOS; detached dashboards not tried on a second display; real sleep/wake,116notification delivery and the launch-at-login switch were not exercised; no VoiceOver review by a person; French not117proofread. One cosmetic issue is known in this build: in the menu bar panel the network line truncates the upload rate118when both rates are wide.119120## 8. Repository121122| Item | Value |123|---|---|124| Repository | <https://git.spboucher.ai/orix-monitor> — public since 2026-11-21 (`gitsrv:srv/git/orix-monitor.git`) |125| Before publication | Private SSH-only repository `gitsrv:srv/git-private/orix-monitor.git`, removed after the public push was verified |126| Branch | `main` |127| Tag `v1.0.0` | commit `2e9d6ed2156f72fa7f7603c5fc14125d08939e58` (the source of the notarized build); later commits on `main` are documentation |128| Committed | Source, tests, scripts, documentation, screenshots, icon artwork. **Not** committed: binaries (`dist/`), build products, the generated `.xcodeproj`, notary responses, any credential |129130## 9. Signing, notarization, Gatekeeper131132| Item | Value |133|---|---|134| Version / build | 1.0.0 / 1 |135| Signing identity | Developer ID Application: Simon-Pierre Boucher (3YM54G49SN) |136| Hardened Runtime | Enabled · secure timestamp present · entitlements: none · `get-task-allow`: absent |137| Notarization, app | Submission `4f0026aa-6755-471b-9392-b06671e90083` — **Accepted** |138| Notarization, disk image | Submission `13bf43c8-e0e8-4c84-acfc-8e74bc3f6f23` — **Accepted** |139| Stapling | App and DMG stapled; `stapler validate` succeeded on both |140| Gatekeeper, app | `accepted`, source = Notarized Developer ID |141| Gatekeeper, DMG | `accepted`, source = Notarized Developer ID |142| `ORIX-Monitor-1.0.0.dmg` | 3,455,660 bytes · SHA-256 `76ae76570e2c7654cdf377922bc85967d04eb20eedeaabd7675791f5b84e3be2` |143| `ORIX-Monitor-1.0.0.zip` | 2,515,813 bytes · SHA-256 `3b91a67ff51dd880bc1250491b1abc6591a95349717318139571023bcf1ad752` |144| SPB Git release URL | <https://git.spboucher.ai/orix-monitor/releases> (tag `v1.0.0`) |145| Download URLs | `https://git.spboucher.ai/releases/orix-monitor/v1.0.0/ORIX-Monitor-1.0.0.dmg` · `…/ORIX-Monitor-1.0.0.zip` · `…/SHA256SUMS.txt` · `…/RELEASE-NOTES.md` |146| Download verification | Fetched from the public URLs on 2026-11-21: HTTP 200, 3,455,660 and 2,515,813 bytes, both SHA-256 match, `stapler validate` succeeded, `spctl` accepted the downloaded DMG (Notarized Developer ID) |147148Apple's notarization requirements were read from Apple's documentation on 2026-11-21. The Apple Developer account pages and149App Store Connect were not opened: nothing here needs them and they require the owner's sign-in. Four earlier accepted150submissions belong to superseded builds of the same version (`docs/release/notarization.md`); their artifacts were deleted.151152Accessibility: Apple's automated audit recorded 138 contrast notes, 3 near-passes, 136 parent/child notes, 100 unknown-role153notes and 63 missing-description notes across 12 pages in Dark Mode. Text colour tokens were verified numerically at 4.5:1154or better; the audit does not say which elements its contrast notes concern, and they remain unexplained.155156## 10. Outstanding approvals157158None. Publication of the source and of release v1.0.0 was authorised by the owner on 2026-11-21 and carried out.159160One thing to know: the screenshots, now public, show the names of processes and volumes that were on the development Mac at161capture time (addresses and user names are masked). They can be recaptured with `Scripts/screenshots.sh` if any of them162should not be visible.163164## 11. Remaining work165166- Find the cause of the graphics-memory swing; measure energy impact and the menu-bar-only state.167- VoiceOver review by a person; rerun the audit after hiding decorative drawings; audit Light Mode.168- Test on macOS 15 and on a Mac with a second display; exercise real sleep/wake, notifications and launch at login.169- Proofread the French; widen the menu bar panel's network line.170- Automatic updates once a release feed exists; an editor for insight thresholds; per-core history.171
Files main
Menu

Theme