[JarryShaw]

Please follow the guide below

• You will be asked some questions, please read them carefully and answer honestly

• Put an x into all the boxes [ ] relevant to your pull request (like that [x])

• Use Preview tab to see how your pull request will actually look like

• Searched for similar pull requests

• Followed the coding style (make pylint, make mypy, make isort)

• make test passes, and a test case covers the change. Narrow runs only, listed below.

• Added a changelog entry under docs/source/changelog/ and regenerated CHANGELOG.md, if the change is user-visible: 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?

Tick the commit type your subject line carries.

• fix — corrects a defect
• feat — adds a feature
• perf — changes performance, not behaviour
• refactor — changes neither behaviour nor performance
• test — tests only
• docs — documentation only
• ci — workflows or build tooling
• release — bumps the version or rolls up a distribution
• chore — anything else

---

Description of your pull request and other information

Closes #1709. This applies the owner's rulings on #1701 and #1703 to the six stdlib feature probes left in pcapkit/.

• compat re-exports: get_annotations (3.10+, otherwise a class's own __dict__['__annotations__']) and UnionType (3.10+, otherwise None), both behind sys.version_info. infoclass and protocol import both from there, and _own_annotations is gone.
• MultiInfo / OrderedMultiInfo: they bind __reversed__ behind sys.version_info >= (3, 8) instead of hasattr(dict, '__reversed__').
• _make_index_calls: it reads an ast call's end_lineno behind sys.version_info >= (3, 8) instead of through getattr.
• New guard: tests/project/test_stdlib_feature_probes_1709_unit.py scans pcapkit/ for literal-name getattr/hasattr calls on a module or builtin class, or for end_lineno/end_col_offset on any object. It fails on main and on round 1 of this PR.
• The other 224 literal-name calls read optional protocol fields, pcapkit's own attributes, third-party objects, or stdlib attributes every supported Python has.
• Behaviour: unchanged. Dumps of 492 Info classes are byte-identical before and after on 3.8, 3.9, 3.10 and 3.14.
• Conflict with refactor(corekit)!: move chardet into corekit (#1712) #1729: both edit protocol.py's import lines. Whichever merges second gets rebased.

Tests: the compat, infoclass, probe-scan, round-trip and protocol-make selections pass on 3.8, 3.9, 3.10 and 3.14.

[JarryShaw]

Verdict: GOOD TO GO at cfb909b52. Sonnet cross-reviews (Opus wrote the change) checked two rounds.

• Round 1 missed one probe: getattr(node, 'end_lineno', node.lineno) at protocol.py:97. Round 2 gates it, and the widened scan catches the old line.
• Gates: they select exactly the old behaviour on 3.8, 3.9, 3.10 and 3.14. Old and new readers agree on 7 class shapes, and the 492-class dumps are byte-identical.
• compat.__all__: each name appears once. The line-number comment in test_compat.py still matches the file.
• Enumeration: I independently listed every literal-name getattr/hasattr call. None is a stdlib version probe.

[github-actions]

Coverage: 89.35% (unit tier, Python 3.15, cfb909b52, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	3	20	3	95.16%
pcapkit/const	18860	954	2368	868	91.06%
pcapkit/corekit	2374	67	762	35	96.49%
pcapkit/dumpkit	275	3	100	3	98.40%
pcapkit/foundation	2885	144	1010	34	94.20%
pcapkit/interface	138	7	52	5	93.68%
pcapkit/protocols	17170	225	4586	189	98.08%
pcapkit/toolkit	1148	290	450	10	71.21%
pcapkit/utilities	515	4	158	4	98.81%
pcapkit/vendor	4417	2342	1010	157	43.38%

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