[JarryShaw]

• Searched for similar pull requests

• Followed the coding style (make pylint, make mypy, make isort): ran mypy and pylint on the three IPX files only; mypy is clean, and pylint adds only the existing stub-signature warnings

• make test passes, and a test case covers the change: ran the modules listed below

• Changelog entry: N/A, centralised in docs(changelog): shared 1.5.0 changelog — long-lived, merges last (#610, #616, #617, #618, #620) #657

• fix: corrects a defect

---

Closes #1433

This is the IPX counterpart of #1431 and has the same shape, without depending on it. The schema gains a length-less trailer PayloadField after the payload. read() keeps it as info.trailer only when it is non-empty. make(trailer=) writes it after the packet and leaves it out of the Packet Length, and _make_data round-trips it. Breaking: info gains a key on padded frames.

case	before	after
Ethernet + IPX(30), zero / junk padding to 60	60 → 44	60 → 60, trailer 16 B
bare IPX(34) + 5 extra	39 → 34	39 → 39
exact / truncated (Length 40, 34 captured)	34 → 34	unchanged, no trailer

Tests: the new tests/protocols/internet/test_ipx_trailer_kept_unit.py passes 4 tests and 4 subtests. Without the fix, 2 tests and 2 subtests fail. The other IPX modules, every tests/protocols/link module, tests/project (404 passed, 1 skipped) and the protocols/internet leg (580 tests, OK) all pass. Per-frame info for every examples/captures/ file is identical before and after.

[JarryShaw]

Verdict: GOOD TO GO at 3075bd3d2 (Sonnet cross-review; the author ran on Opus).

• Re-derived by me: a well-formed 30-octet IPX header in a 60-octet Ethernet frame rebuilds at 44 octets on main (a8cc02594) and stays exact at 60 here.
• Reviewer:
  • Junk padding: also 60 → 60, with a 16 B trailer.
  • Bare IPX with Length 34 plus 5 extra octets: 39 → 34 on main, exact here.
  • Exact-length and truncated packets (Length 40, 34 captured) have no trailer key and no over-read.
  • Length below the header (20, 0, 29): unchanged from main, and exact.
• Payload to the next layer is exactly Length − 30 (Raw b'abcd' for Length 34).
• make(trailer=): the trailer is written after the packet and Length excludes it. Make → parse round-trips.
• Same shape as fix(ipv4,ipv6)!: keep the octets past Total Length or Payload Length as captured (#1209) #1431: a PayloadField(default=b'') after the payload, if schema.trailer: in read, and data.get('trailer', b'') in _make_data. None handling is the same in both, and so is the bytearray rejection, which is shared and pre-existing. The two PRs touch disjoint files.
• Tests:
  • With the 3 IPX files reverted, 4 tests fail.
  • test_ipx_unit, test_ipx_ah_roundtrip_unit, tests/protocols/link (70), test_docstring_contract and tests/project (404) all pass.
  • Per-frame info is identical across the 22 sample captures. None of them carries IPX.
• ! is justified, on the same grounds as fix(ipv4,ipv6)!: keep the octets past Total Length or Payload Length as captured (#1209) #1431.

[github-actions]

Coverage: 89.45% (unit tier, Python 3.14, 3075bd3d2, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	4	20	4	93.55%
pcapkit/const	18791	968	2352	869	90.90%
pcapkit/corekit	2038	89	634	25	94.76%
pcapkit/dumpkit	234	2	84	2	98.74%
pcapkit/foundation	2637	98	928	48	94.78%
pcapkit/interface	112	7	40	5	92.11%
pcapkit/protocols	16618	221	4476	185	98.04%
pcapkit/toolkit	539	74	168	3	84.58%
pcapkit/utilities	429	4	122	4	98.55%
pcapkit/vendor	4409	2342	1006	157	43.25%

Per-file detail: the coverage-html artifact of this run.