[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

---

Summary

MANIFEST.in had lines that matched nothing, as #1794 measured. All are removed; none needed correcting.

• global-include *.rst: every tracked .rst is under docs/, which prune docs drops.
• include pcapkit/_extern: include matches files, not a directory. The conda recipe builds the wheel straight from the tree with pip install ., never an sdist, and the vendored modules reach the wheel through package discovery either way.
• exclude pcapkit/protocols/*/NotImplemented: the placeholders are directories, so the pattern matched nothing. They ship either way, as documentation.rst says.
• Evidence: plain sdist and wheel listings are identical before and after, and both build warnings are gone. With a conda-style _extern, only two vendored .rst files that nothing reads drop out.
• Test: tests/project/test_build_leftovers_1773.py gains a guard that every non-prune directive changes the selection; it fails on main on exactly these lines.

Closes #1794

[JarryShaw]

Verdict: GOOD TO GO at 6d63caaa3. A Sonnet cross-review (author: Opus) ran its own sdist and wheel builds, with and without a conda-style _extern, and confirmed the author's listings and that both warnings are gone. The new guard fails on exactly these lines and catches added no-op lines. The review's one wording point, a MANIFEST.in comment claim, is fixed at this head (comment-only amend, verified).

[github-actions]

Coverage: 89.60% (unit tier, Python 3.15, 6d63caaa3, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	3	20	3	95.16%
pcapkit/const	18862	954	2368	868	91.06%
pcapkit/corekit	2462	69	780	35	96.55%
pcapkit/dumpkit	276	3	100	3	98.40%
pcapkit/foundation	2915	164	1016	34	93.59%
pcapkit/interface	138	7	52	5	93.68%
pcapkit/protocols	17197	224	4580	189	98.09%
pcapkit/toolkit	872	177	298	8	77.18%
pcapkit/utilities	512	2	158	4	99.10%
pcapkit/vendor	4417	2342	1010	157	43.38%

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

[JarryShaw]

_extern are for Conda distros iirc.

[JarryShaw]

Yes, pcapkit/_extern is the conda recipe's vendor directory: conda/pypcapkit/build.sh and bld.bat fill it from conda/requirements.txt. The removed line never shipped it, though. I re-measured just now with setuptools' own FileList on a tree that has pcapkit/_extern/certifi/{__init__.py,cacert.pem}:

• include pcapkit/_extern matches nothing (warning: no files found matching 'pcapkit/_extern'), because include matches file patterns, not directories.
• recursive-include pcapkit/_extern * picks up all three files, including cacert.pem.

So conda builds today get the vendored .py modules through package discovery and none of their data files. #1814 tracks that gap. This PR only drops the dead line and records why, so #1814's fix (most likely the recursive-include form, or package-data) can replace it. If you would rather keep a placeholder line here until then, say so and I will restore it.