[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?

• docs — documentation only

---

Description

Part of #1103, ruling 1: test narratives in tests/protocols/application/, tests/protocols/transport/ and tests/vendor/ now state the behaviour, not how a change produced it. This is the last of the seven slices.

• Rewrites:
  • "now" and "used to" become timeless.
  • Old-tree proofs ("Pre-fix this packed 11 …") become the defect each test guards against.
  • PR-history module docstrings now describe current behaviour. This covers MPTCP join/capable/subtype/length, the vendor snapshot-restore and destination-path tests, and the HTTP tests.
  • Counts and file:line anchors in rewritten prose are dropped.
• Stale claims corrected against the code:
  • The HTTP/2 preface is skipped under read(version=2) (fix(http): the HTTP/2 connection preface is lost on rebuild #1242), not refused.
  • MP_CAPABLE uses > 12, not == 20.
  • A note claiming a failing subTest reports its parent as passed now matches what pytest 9 does: the failing subtest is a counted SUBFAILED entry, and the parent counts as passed.
• Kept:
  • CPython, IANA and Wikipedia history.
  • Version-pinned measurements.
  • "used to" in the sense of "used for".
  • Change-relative string literals and identifiers, which are code.
• Scope: comments and docstrings only. With docstrings stripped, every changed file has main's AST.
• Tests run: application (244), transport (243), vendor (169), and the tier guard with the docstring contract (140) pass.
• Overlap: fix(tcp): keep Quick-Start rate an int and a Fast Open cookie request bytes (#2167) #2185 shares test_tcp_udp_unit.py and merges cleanly.

[JarryShaw]

Review: NEEDS CHANGES at 530d068ea (Sonnet).

• False present tense: one rewrite says MPTCPFastclose packs 11 octets with no reserved field. It has the reserved field.
• Change-relative forms left behind: "Pre-MPTCP.subtype is declared only under TYPE_CHECKING, so every subtype built via TCP's convenience constructor raises AttributeError #566/TCP._make_mptcp_capable writes length 20/32 where RFC 8684 gives 12/20, and now packs wrong bytes rather than crashing #567", "Before the guard", "formerly rejected", "revert Constructing an MP_JOIN option raises AttributeError: TCP._make builds the options before assigning self._flags #587's hoist", and a "now in the citation".
• Everything else checks out: the corrected claims (fix(http): the HTTP/2 connection preface is lost on rebuild #1242 preface, MP_CAPABLE > 12, the pytest 9 subtest note) hold against the code, AST is clean, and the tests pass. The worker is fixing the rest.

[JarryShaw]

Review fixes pushed: 530d068ea → 23d10991c, rebased onto main.

• MPTCPFastclose: no longer described as lacking a reserved field.
• Change-relative forms: the "Pre-MPTCP.subtype is declared only under TYPE_CHECKING, so every subtype built via TCP's convenience constructor raises AttributeError #566/TCP._make_mptcp_capable writes length 20/32 where RFC 8684 gives 12/20, and now packs wrong bytes rather than crashing #567", "Before the guard", "formerly" and "revert Constructing an MP_JOIN option raises AttributeError: TCP._make builds the options before assigning self._flags #587's hoist" passages now state what each test guards against.
• _guess_version: one line of the decision reason is kept.
• Short-payload docstring: now says ProtocolError (httpv2: the frame guard tests the declared length, not the buffer, so a 4-octet frame can report length=16777215 #799); verified by running it.

AST is clean and all three directories pass (244 / 245 / 169). A delta review is running.

[JarryShaw]

Review: GOOD TO GO at 23d10991c (Sonnet delta). Both required items and both optional items are fixed, each checked against the code, and only the intended hunks moved. The remaining grep hits are all string literals, identifiers or CPython behaviour. The AST check is clean on all 54 files, and application, transport and vendor tests pass.

[JarryShaw]

Ready to merge at 23d10991c. Part of #1103 (ruling 1).

• Review: Sonnet GOOD TO GO, after one NEEDS CHANGES round.
• CI: all checks complete, Required checks passed SUCCESS, 0 failed.
• KTLO: 12 legs, no warnings.
• Ordering: each leg was merged into main and run. protocols: 1923 tests, 0 failures. vendor: 401 tests, 0 failures.
• Conflicts: none against the current main.

[github-actions]

Coverage: 91.55% (unit tier, Python 3.15, 23d10991c, 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	2589	74	816	37	96.51%
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	17433	219	4584	185	98.15%
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.