research: an OTN line port needs no new port type - four specs already file them as Ethernet cages #1938

Merged
jaylouisw merged 1 commit from catalogue/1118-0426 into main 2026-11-18 04:41:30 +00:00
Owner

The correction

research/optixtrans-board-census.md section 2 concluded that #807, #808 and #810 were blocked on two fields, the second being "a port type that can represent an OTN line port", which it said required an engine change. The measurement in that section is right — engine/media.js has no otn or wdm type, and CAGE_MEDIA['sfp+'] offers 10G Ethernet media, so a WDM-side wavelength filed as ethernet can be joined to a 10GBASE-LR peer. The conclusion does not hold.

Four specs on main already file exactly those ports the same way, and all four pass scripts/audit-port-types.js and test/hardware-invariants.test.js:

spec line ports filed as
devices/huawei/optixtrans-e6616.json 28 line* ethernet / cfp2 / 200G
devices/ciena/waveserver-ai.json 6 line* ethernet / qsfp28 / 100G
devices/adva/fsp-3000.json 16 line* ethernet / qsfp28 / 100G
devices/packetlight/pl-2000m.json 1 line* ethernet / qsfp28 / 100G

None of those ports carries Ethernet — a coherent wavelength on an FSP 3000 line card is an OTUk signal, and optixtrans-e6616 is a sibling frame of the queued ones, in the same vendor and the same document family, saying so in notes. So the missing type is an improvement waiting for an engine PR, not a prerequisite for these specs.

What is in the diff

  • research/otn-line-ports-need-no-new-type.md — the convention stated once (cage as built, type: ethernet as the engine's only vocabulary for "framed signal at this rate", the protocol mismatch written into notes, fabric.excludes: ["line*"] so a line port is not double-counted), the exact sentences each sibling uses to carry the compromise, what an otn type would buy and the mechanical migration if it lands, and the label each queued frame will take once its census arrives.
  • research/optixtrans-board-census.md — a dated correction block under section 2, original text kept for its measurement, and the closing list narrowed from two blocked fields to one.

Effect on the queue

The blocker for the three OptiXtrans frames is now singular: one TNG1N* / TNG1M* client-board chapter that states both its front panel and which subrack accepts it. The frame arithmetic and the eventual labels are already determined; the validity of the board pairing is the part no anonymously readable vendor document states.

No issue is answered by this PR and none should change state: the open-request ledger in research/open-device-requests.json is untouched and still accurate.

Gates

  • node scripts/syntax-check.js — 405 files, all OK
  • node scripts/audit-port-types.js — no port contradicts its declared type
  • node scripts/check-control-bytes.js — no control bytes
  • node test/request-queue.test.js — 32 passed
  • node test/apply-research.test.js — 10 passed
  • node test/hardware-invariants.test.js — 26 passed
  • node scripts/regen-registry-doc.js — no diff (2,353 models, 258 vendors)
## The correction `research/optixtrans-board-census.md` section 2 concluded that #807, #808 and #810 were blocked on **two** fields, the second being "a port type that can represent an OTN line port", which it said required an engine change. The measurement in that section is right — `engine/media.js` has no `otn` or `wdm` type, and `CAGE_MEDIA['sfp+']` offers 10G Ethernet media, so a WDM-side wavelength filed as `ethernet` can be joined to a 10GBASE-LR peer. The conclusion does not hold. Four specs on `main` already file exactly those ports the same way, and all four pass `scripts/audit-port-types.js` and `test/hardware-invariants.test.js`: | spec | line ports | filed as | |--|--|--| | `devices/huawei/optixtrans-e6616.json` | 28 `line*` | `ethernet` / `cfp2` / 200G | | `devices/ciena/waveserver-ai.json` | 6 `line*` | `ethernet` / `qsfp28` / 100G | | `devices/adva/fsp-3000.json` | 16 `line*` | `ethernet` / `qsfp28` / 100G | | `devices/packetlight/pl-2000m.json` | 1 `line*` | `ethernet` / `qsfp28` / 100G | None of those ports carries Ethernet — a coherent wavelength on an FSP 3000 line card is an OTUk signal, and `optixtrans-e6616` is a sibling frame of the queued ones, in the same vendor and the same document family, saying so in `notes`. So the missing type is an improvement waiting for an engine PR, not a prerequisite for these specs. ## What is in the diff - `research/otn-line-ports-need-no-new-type.md` — the convention stated once (cage as built, `type: ethernet` as the engine's only vocabulary for "framed signal at this rate", the protocol mismatch written into `notes`, `fabric.excludes: ["line*"]` so a line port is not double-counted), the exact sentences each sibling uses to carry the compromise, what an `otn` type would buy and the mechanical migration if it lands, and the label each queued frame will take once its census arrives. - `research/optixtrans-board-census.md` — a dated correction block under section 2, original text kept for its measurement, and the closing list narrowed from two blocked fields to one. ## Effect on the queue The blocker for the three OptiXtrans frames is now singular: **one `TNG1N*` / `TNG1M*` client-board chapter that states both its front panel and which subrack accepts it**. The frame arithmetic and the eventual labels are already determined; the validity of the board pairing is the part no anonymously readable vendor document states. No issue is answered by this PR and none should change state: the open-request ledger in `research/open-device-requests.json` is untouched and still accurate. ## Gates - `node scripts/syntax-check.js` — 405 files, all OK - `node scripts/audit-port-types.js` — no port contradicts its declared type - `node scripts/check-control-bytes.js` — no control bytes - `node test/request-queue.test.js` — 32 passed - `node test/apply-research.test.js` — 10 passed - `node test/hardware-invariants.test.js` — 26 passed - `node scripts/regen-registry-doc.js` — no diff (2,353 models, 258 vendors)
research: an OTN line port needs no new port type - four specs already file them as Ethernet cages
Some checks failed
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 4s
CI / Classify the change (pull_request) Successful in 6s
CI / Docker build (pull_request) Successful in 22s
CI / Test (Node 20) (pull_request) Successful in 46s
CI / Test (Node 18) (pull_request) Successful in 55s
CI / Test (Node 22) (pull_request) Successful in 1m3s
CI / Server smoke test (pull_request) Failing after 12m14s
CI / Arm auto-merge (pull_request) Has been cancelled
78d2f2638e
The E9600 blocker note claimed the family was stuck on two fields, the second
being 'a port type that can represent an OTN line port'. The measurement is
right (no otn/wdm type exists) but the conclusion is wrong: optixtrans-e6616
(28 x 200G line as ethernet/cfp2), ciena/waveserver-ai, adva/fsp-3000 and
packetlight/pl-2000m all ship OTN and WDM line ports as Ethernet cages with the
compromise in notes, and all four pass audit-port-types. The missing type is an
engine improvement, not a gate - so the family is blocked on the board census
alone.
jaylouisw force-pushed catalogue/1118-0426 from 78d2f2638e
Some checks failed
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 4s
CI / Classify the change (pull_request) Successful in 6s
CI / Docker build (pull_request) Successful in 22s
CI / Test (Node 20) (pull_request) Successful in 46s
CI / Test (Node 18) (pull_request) Successful in 55s
CI / Test (Node 22) (pull_request) Successful in 1m3s
CI / Server smoke test (pull_request) Failing after 12m14s
CI / Arm auto-merge (pull_request) Has been cancelled
to cba620f374
All checks were successful
CI / Classify the change (pull_request) Successful in 6s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 5s
CI / Server smoke test (pull_request) Successful in 8s
CI / Docker build (pull_request) Successful in 23s
CI / Test (Node 20) (pull_request) Successful in 53s
CI / Test (Node 18) (pull_request) Successful in 1m0s
CI / Test (Node 22) (pull_request) Successful in 58s
CI / Arm auto-merge (pull_request) Successful in 8s
2026-11-18 04:39:39 +00:00
Compare
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!1938
No description provided.