[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, 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

Before 3.10 (bpo-43901), a class with no annotations of its own returns its nearest base's from cls.__annotations__. The new _own_annotations reads the class's own __dict__ below 3.10 and uses inspect.get_annotations from 3.10 on. It cannot use __dict__ alone there, because 3.14 keeps lazily evaluated annotations out of __dict__; this is the same split as ProtocolBase in protocols/protocol.py. info_final and _nested_types both call it now.

• A field-less subclass, and a field-less subclass of that, now get (self, dict_=None, **kwargs) on every version. A diamond whose middle class is field-less now gets (self, a, c, d); before this fix 3.8 and 3.9 gave (self, c, a, d).
• from_dict now rebuilds an Info field that a field-less subclass inherits. Before, 3.8 and 3.9 resolved that field's annotation in the subclass's namespace and skipped it.
• Across pcapkit's 494 Info classes, the old and new readers differ only for the four _MultiInfo classes, and those generate no __init__. So no shipped class changes.
• The four cls.__annotations__[...] writes (protocols/data/internet/ipv4.py:74, :150, :345, ipv6.py:76) were measured on 3.8, 3.9, 3.10 and 3.14. Each target class declares its own annotations, and nothing leaked into any base. They are unchanged.

Tests: the three tests from the issue and the new OwnAnnotationsTests fail on origin/main under 3.8 and 3.9. With this fix they pass on 3.8, 3.9, 3.10 and 3.14, under both pytest and unittest. 15 related modules pass on every version. The one exception is test_final_enforcement.py::VersionPortabilityTests on 3.8, which already fails on main (ast.unparse). infoclass.py coverage from this test module goes from 66% to 72% on 3.14.

Closes #1692

[JarryShaw]

Verdict: GOOD TO GO at 67ee18a25. A Sonnet cross-review reproduced the three failures from #1692 on 3.8 and 3.9 at base. On this head they pass on 3.8, 3.9, 3.10 and 3.14, with 23 OK under unittest. The new OwnAnnotationsTests fail against the base on 3.8 and 3.9. Mutants that revert only info_final, or only _nested_types, each fail their own tests. No regression was found on 3.10 or 3.14.

[github-actions]

Coverage: 89.47% (unit tier, Python 3.15, 67ee18a25, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	3	20	3	95.16%
pcapkit/const	18858	954	2366	868	91.06%
pcapkit/corekit	2372	67	762	35	96.49%
pcapkit/dumpkit	275	3	100	3	98.40%
pcapkit/foundation	3035	151	1076	34	94.28%
pcapkit/interface	138	7	52	5	93.68%
pcapkit/protocols	17166	226	4580	190	98.07%
pcapkit/toolkit	1437	279	592	5	78.12%
pcapkit/utilities	494	4	146	4	98.75%
pcapkit/vendor	4417	2342	1010	157	43.38%

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