[JarryShaw]

• Searched for similar pull requests
• Followed the coding style
• make test passes, and a test case covers the change
• 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?

• fix — bug fix

---

Description

Closes #2280.

• Fix: _make_opt_lmaa's code now defaults from the address family (IPv4 → IPv4_LMAA, IPv6 → IPv6_LMAA) instead of always defaulting to IPv6_LMAA; an explicit code that disagrees with the family raises ProtocolError. The option= rebuild path is unchanged, so a captured mismatch still rebuilds byte-exact.
• Fail-without-fix: 6 failed on origin/main (two are by-name regression guards the amendment added). Passing on the head.

[JarryShaw]

Review: NEEDS CHANGES at 269cdadeb (Sonnet) — a real regression.

• Regression: _make_opt_lmaa's default changed from an enum (LMAAddressCode.IPv6_LMAA) to None, but _make_index's str-name resolution needs an enum default to resolve against. code='IPv4_LMAA' or 'IPv6_LMAA' now raises ProtocolNotImplemented where origin/main built it; every sibling _make_opt_* keeps an enum default for exactly this reason (per test_make_index_str_unit.py).
• Fix: pass namespace=code_namespace if code_namespace is not None else LMAAddressCode to the _make_index call, which the reviewer verified restores string-name resolution without breaking the new mismatch check.
• Confirmed otherwise: the family-derived default, the enum/int mismatch rejection, and the option= byte-exact rebuild path all hold. Fail-without-fix: 4 failed on origin/main. On the head, 188+197 passed.
• Non-blocking: add code='IPv4_LMAA'/'IPv6_LMAA' subtests so this class of regression is caught; a stale "measured without the fix" docstring figure in test_mh_unit.py is now wrong and should be updated.

[JarryShaw]

Amended 269cdadeb → 392ea7d8a: fixed the regression. _make_index's namespace is now code_namespace if code_namespace is not None else LMAAddressCode, so a str code name resolves again. Added code='IPv4_LMAA'/'IPv6_LMAA' subtests to both the matching and contradicting loops in test_mh_lmaa_code_follows_the_address_family — they fail on origin/main (6 failed total with the full copied-in test file) and pass on the new head (68 passed across the MH, _make_index str-resolution and docstring-contract tests).

[JarryShaw]

Review: GOOD TO GO at 392ea7d8a (Sonnet, round 2).

• Confirmed: code='IPv4_LMAA'/'IPv6_LMAA' now build byte-identical to their enum/int equivalents on both families; a contradicting name still raises ProtocolError with the same message form, not ProtocolNotImplemented; no other caller passes code_namespace to this method, so the fallback is safe; round 1's claims still hold (306-case matrix, 28 of which are the new by-name cases); option= rebuild is untouched.
• Fail-without-fix: reverting just the mh.py hunk fails the two new by-name regression subtests (ProtocolNotImplemented). On origin/main, the full test file now fails 6 (the two by-name "matching" subtests pass there, since they're regression guards, not fail-on-main proof).
• Non-blocking: the PR body's "4 failed on main" is stale (now 6, from the added subtests); a same-class str-name gap exists elsewhere (hopopt/ipv6_opts._make_opt_mpl, _make_opt_smf_dpd, httpv1._make_start_line) — worth its own issue.

[github-actions]

Coverage: 91.58% (unit tier, Python 3.15, 392ea7d8a, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	4	20	2	95.16%
pcapkit/const	19599	881	2618	867	92.11%
pcapkit/corekit	2596	74	818	37	96.51%
pcapkit/dumpkit	282	3	102	3	98.44%
pcapkit/foundation	3029	23	1050	25	98.82%
pcapkit/interface	138	4	52	3	96.32%
pcapkit/protocols	17449	218	4592	185	98.15%
pcapkit/toolkit	914	36	334	10	95.51%
pcapkit/utilities	502	7	144	5	98.14%
pcapkit/vendor	4432	1988	1018	167	51.87%

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

[JarryShaw]

Ready to merge at 392ea7d8a. Closes #2280.

• Review: GOOD TO GO (Sonnet), two rounds — round 1 caught a real regression (a str code name broke), fixed and re-verified.
• CI: Required checks passed SUCCESS, 0 failures, KTLO clean.
• Ordering leg: protocols gave 1962 tests with 0 failures, run with the PR merged into current main.
• Merge state: mergeable.