[JarryShaw]

• Searched for similar pull requests
• Followed the coding style (make pylint, make mypy, make isort) — N/A, .rst-only diff
• make test passes, and a test case covers the change — N/A, docs only; tests/project passes
• 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?

• docs — documentation only

---

Description of your pull request and other information

Slice of #719: docs/source/pcapkit/** except protocols/ (18 files, +143 / -159). Cuts host-specific "measured on" logs, "has since landed" / "was added" / "now" prose; keeps every rationale. Fixes: engine constraint claim (PyPCAP has no Python-ceiling check), interface table missing 3 engine constants, typing.final gate 3.11 in the compat table, heading typos.

Not touched: corekit/, dumpkit/, toolkit/, utilities/logging.rst, foundation/registry.rst, engines/builtin.rst, engines/engine.rst, most const/ and vendor/ pages (generated autodoc, nothing to cut), and the TCP/IPv6 reassembly data-model blocks beyond the two timed phrases. The slice was not fully rewritten.

Checks: sphinx-build -b dummy -n fresh dirs, branch vs main, slice warnings 47 vs 47, none new (1301 vs 1301 total). tests/project 274 passed.

[JarryShaw]

Cross-review verdict on cdb7f6965: NEEDS CHANGES (ran on Opus; authored on Sonnet). Two sentences in the rewritten engines paragraph are wrong against the code, and one PR-body tickbox overclaims.

1. foundation/engines/index.rst:142-143 omits PyShark's tshark check. PyShark.unsupported_reason checks its Python ceiling and whether get_process_path() finds tshark (pyshark.py:126, :134-144). The list reads as complete, so a reader would conclude a missing tshark goes uncaught.
2. :144-145 gives the opposite of PyPCAP's real reason. The page says PyPCAP has no Python ceiling "since upstream cannot be installed above it". The docstring (pypcap.py:148-151) says it is deliberately not a ceiling even though upstream cannot install on 3.12+, because the check is which distribution is present: a version check would refuse a working build. docs: sweep all prose for concision and accuracy, and drop timed context outside the changelog #719 keeps design reasons, so use the real one or drop the clause.
3. PR body: the "make test passes, and a test case covers the change" box should read "N/A — docs only; tests/project passes". No test was added and only tests/project ran. The same applies to the lint and type-check box.

Confirmed:

• Removing "every engine overrides unsupported_reason" is right: only PyShark, PyPCAP, PCAP_CT and PyPCAPFile define one.
• The three new interface rows render: the table goes from 18 to 21 rows.
• typing.final at 3.11 is right. compat.py:157-174 gates on 3.11, the first release where final sets __final__.
• No reference used any of the retitled headings.
• The reviewer's own Sphinx -n build matches main exactly, warning for warning.

Found along the way, now #1045: a missing engine produces two EngineWarnings.

[JarryShaw]

Cross-review verdict on 1959303a1: GOOD TO GO (ran on Sonnet).

Every claim in the new engines/index.rst sentence checks out against origin/main:

• PyShark checks a 3.14 ceiling, then tshark (pyshark.py:126, :134-151).
• PyPCAPFile checks only its 3.12 ceiling.
• PCAP_CT checks for a system libpcap.
• PyPCAP and PCAP_CT both call _pcap_backend.wrong_backend_reason.
• PyPCAP's deliberate absence of a ceiling check, and the reason for it, match pypcap.py:146-151.

No other engine overrides unsupported_reason. The paragraph agrees with the version table and with the later bullet. The make test and coding-style boxes are now N/A with reasons.

Nits, not blocking: "a fall back" would read better as "a fallback". The source qualifies the refused build as hypothetical; the page does not.