devices(poe): four budgets quoted only in prose, and a GS110TP with eight unpowered ports #2287

Merged
jaylouisw merged 1 commit from catalogue/1120-0922 into main 2026-11-20 09:38:51 +00:00
Owner

Neither open device request is answered here — refs #130 and #803 only; both remain blocked on vendor material that does not exist (see docs/DEVICE_REGISTRY.md). This is the catalogue increment for the run.

The defect

node scripts/audit-poe-budget.js separates two failures. The loud one is a budget with no PSE port. The quiet one, closed here, is a spec whose description/notes states the total wattage while the structured poe block is absent: a human reads the number, engine/simulator.js hits if (!dev.poeBudget) continue;, and show power inline answers "budget not modelled" for a box the file can quote to the watt.

75 of 411 PSEs cannot be asked how much they can supply; 10 of those already quote a figure in prose
→ 71 of 411; 6 in prose.

Changed (each figure from the vendor's own document, URL recorded in poe.sourceUrl)

spec figure source
devices/tp-link/tl-sg3428xmp.json 384 W total, 30 W/port https://www.tp-link.com/us/business-networking/omada-switch-poe/tl-sg3428xmp/ (200, 189,932 B)
devices/tp-link/tl-sg2428p.json 250 W total https://www.tp-link.com/us/business-networking/omada-switch-poe/tl-sg2428p/ (200, 181,862 B)
devices/netgear/gs110tp.json 46 W total, 15.4 W/port https://www.netgear.com/images/datasheet/switches/SmartSwitches/GS110TP.pdf (200, 405,458 B, 3 pp)
devices/netgear/gs116lp.json 76 W (FlexPoE 115/183 W with a separate supply) https://www.netgear.com/business/wired/switches/unmanaged/gs116lp/ (200, 675,591 B)

GS110TP was worse than a missing number. Its eight copper ports carried no poeOut at all and poe-pse/802.3af were missing from capabilities, while its description promised 802.3af PoE on all eight: a silent PSE, the same class as the EdgePoint S16. Ports 1-8 now carry poeOut: 15.4 / poeOutClass: "PoE", and 8 x 15.4 W = 123 W of ceiling against a 46 W supply is stated in poe.note.

Both TP-Link budgets are far under their port ceilings (24 x 30 W = 720 W) and carry the vendor's "based on laboratory testing, not guaranteed" footnote; that is recorded rather than dropped.

Evidence: research/poe-budget-prose-four-models.md, including the five devices left untouched and why (the Aruba 2540 datasheet host answers 403; the two Hikvision NVR figures are unread this pass).

Gates

  • node scripts/syntax-check.js — 487 files, all OK
  • node scripts/audit-port-types.js — no port contradicts its declared type
  • node scripts/audit-poe-budget.js — 75 → 71 PSEs, 10 → 6 prose; 0 budgets with no PSE
  • node test/hardware-invariants.test.js — 26 passed, 0 failed
  • node test/import-devices-capabilities.test.js — 34 passed, 0 failed
  • node scripts/run-tests.js — 407 suites, 13,513 assertions, 0 failed
  • node scripts/regen-registry-doc.js — catalogue pages current, no diff
Neither open device request is answered here — refs #130 and #803 only; both remain blocked on vendor material that does not exist (see `docs/DEVICE_REGISTRY.md`). This is the catalogue increment for the run. ## The defect `node scripts/audit-poe-budget.js` separates two failures. The loud one is a budget with no PSE port. The quiet one, closed here, is a spec whose `description`/`notes` states the total wattage while the structured `poe` block is absent: a human reads the number, `engine/simulator.js` hits `if (!dev.poeBudget) continue;`, and `show power inline` answers "budget not modelled" for a box the file can quote to the watt. `75 of 411 PSEs cannot be asked how much they can supply; 10 of those already quote a figure in prose` → **`71 of 411`; `6` in prose.** ## Changed (each figure from the vendor's own document, URL recorded in `poe.sourceUrl`) | spec | figure | source | |--|--|--| | `devices/tp-link/tl-sg3428xmp.json` | 384 W total, 30 W/port | <https://www.tp-link.com/us/business-networking/omada-switch-poe/tl-sg3428xmp/> (200, 189,932 B) | | `devices/tp-link/tl-sg2428p.json` | 250 W total | <https://www.tp-link.com/us/business-networking/omada-switch-poe/tl-sg2428p/> (200, 181,862 B) | | `devices/netgear/gs110tp.json` | 46 W total, 15.4 W/port | <https://www.netgear.com/images/datasheet/switches/SmartSwitches/GS110TP.pdf> (200, 405,458 B, 3 pp) | | `devices/netgear/gs116lp.json` | 76 W (FlexPoE 115/183 W with a separate supply) | <https://www.netgear.com/business/wired/switches/unmanaged/gs116lp/> (200, 675,591 B) | **GS110TP was worse than a missing number.** Its eight copper ports carried no `poeOut` at all and `poe-pse`/`802.3af` were missing from `capabilities`, while its description promised 802.3af PoE on all eight: a silent PSE, the same class as the EdgePoint S16. Ports 1-8 now carry `poeOut: 15.4` / `poeOutClass: "PoE"`, and 8 x 15.4 W = 123 W of ceiling against a 46 W supply is stated in `poe.note`. Both TP-Link budgets are far under their port ceilings (24 x 30 W = 720 W) and carry the vendor's "based on laboratory testing, not guaranteed" footnote; that is recorded rather than dropped. Evidence: `research/poe-budget-prose-four-models.md`, including the five devices left untouched and why (the Aruba 2540 datasheet host answers 403; the two Hikvision NVR figures are unread this pass). ## Gates - `node scripts/syntax-check.js` — 487 files, all OK - `node scripts/audit-port-types.js` — no port contradicts its declared type - `node scripts/audit-poe-budget.js` — 75 → 71 PSEs, 10 → 6 prose; 0 budgets with no PSE - `node test/hardware-invariants.test.js` — 26 passed, 0 failed - `node test/import-devices-capabilities.test.js` — 34 passed, 0 failed - `node scripts/run-tests.js` — 407 suites, 13,513 assertions, 0 failed - `node scripts/regen-registry-doc.js` — catalogue pages current, no diff
devices(poe): four budgets quoted in prose but never read by the engine, and a silent 8-port PSE
All checks were successful
CI / Classify the change (pull_request) Successful in 13s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 14s
CI / Test (Node 22) (pull_request) Successful in 54s
CI / Test (Node 20) (pull_request) Successful in 55s
CI / Server smoke and Docker build (pull_request) Successful in 23s
CI / Test (Node 18) (pull_request) Successful in 1m1s
CI / Arm auto-merge (pull_request) Successful in 4s
4cda2ee66c
forgejo-actions scheduled this pull request to auto merge when all checks succeed 2026-11-20 09:32:56 +00:00
jaylouisw deleted branch catalogue/1120-0922 2026-11-20 09:38:56 +00:00
Sign in to join this conversation.
No reviewers
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
jaylouisw/packetforge!2287
No description provided.