[JarryShaw]

• Searched for similar pull requests
• Followed the coding style
• make test passes, and a test case covers the change (prose only; narrow selections below)
• Changelog entry: N/A — centralised in docs(changelog): shared 1.5.0 changelog — long-lived, merges last (#610, #616, #617, #618, #620) #657

What is the purpose of your pull request?

• docs — documentation only

---

Description

Part of #1103, round 5: the link and transport slice, on main 4c3d30163. This PR changes prose only; the AST with docstrings stripped is identical on all 7 changed .py files.

• Registry comments on TCP.__option__, __mp_option__, and SCTP.__chunk__/__parameter__/__cause__ claimed a missing method name makes the value parse by itself. In fact a missing name falls back to *_donone/*_unknown, and a non-string value is a (parser, constructor) tuple. The register_* meth arguments now say so too.
• TCP.__proto__: the comment said HTTP dispatch avoids a trial parse, but HTTP._guess_version trial-parses as a last resort. It also said UDP binds a proxy; UDP binds the same HTTP class.
• ARP._make_proto_resolve: FieldValueError is raised for invalid and wrong-version addresses, not only for bools (checked at runtime).
• L2TP module and l2tp.rst: dropped the wrong HTTP/3–QUIC analogy for L2F/L2TPv1. Link._read_protos returns an EtherType.
• Tests: tests/project/test_docs_* plus test_conventions_doc_claims: 94 passed. tests/protocols/{link,transport} passes once the fixture captures are generated.

[JarryShaw]

GOOD TO GO at 3932865e1 (Sonnet review).

• TCP and SCTP dispatch: read from the code. A string value falls back to *_donone/*_unknown, which exist for every kind, and a non-string value is indexed [0]/[1].
• HTTP._guess_version: it trial-parses only as a fall-through, and UDP binds the same HTTP class.
• ARP: FieldValueError reproduced for invalid, wrong-version and bool addresses.
• L2TP: the HTTP/3–QUIC analogy was wrong, and its replacement is accurate.
• Link._read_protos: returns Enum_EtherType.get(...).
• Checks: no new counts, timed context or file:line anchors; nothing in tests/ pins the changed text. The AST guard passes on all 7 files, and tests/project/test_docs_* plus the conventions test pass (94).
• UNVERIFIED: the new :attr:/:meth: cross-references were not built.

[github-actions]

Coverage: 91.55% (unit tier, Python 3.15, 3932865e1, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	4	20	2	95.16%
pcapkit/const	19599	880	2618	866	92.12%
pcapkit/corekit	2589	75	816	37	96.48%
pcapkit/dumpkit	282	3	102	3	98.44%
pcapkit/foundation	2971	23	1036	25	98.80%
pcapkit/interface	138	4	52	3	96.32%
pcapkit/protocols	17418	219	4574	185	98.14%
pcapkit/toolkit	869	36	298	10	95.20%
pcapkit/utilities	504	7	146	5	98.15%
pcapkit/vendor	4432	1988	1018	167	51.87%

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

[JarryShaw]

Ready to merge at 3932865e1.

• Review: Sonnet GOOD TO GO.
• CI: all 61 CheckRuns passed; none failed or are still running. Required checks passed is SUCCESS, and the KTLO legs raised no warnings.
• Ordering: I merged this PR into main and ran tests/project plus the root modules under the plain unittest ordering leg: 2288 tests, 0 failures, 0 errors.
• Merge conflicts: none with the other round-5 PRs.