[JarryShaw]

• 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
• Added a 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?

• fix — corrects a defect

---

Description of your pull request and other information

Closes #2126.

• Cause: conda-forge's CPython 3.15 on Windows uses CFEP-27's lib/python/site-packages. conda-build 26.9.1 sets SP_DIR from the python package's python_site_packages_path. built_for only knew Lib/site-packages, so it returned python and failed the win-64 3.15 smoke in run 40034513322.
• Fix: built_for recognises both Windows layouts (WINDOWS_SITES) and reads the version from the pyXY build string for either.
• Package: unaffected. conda-build's test: imports: pcapkit passed in a fresh 3.15.1 env.
• Tests: TestTheWindowsLayout covers the CFEP-27 path in built_for and end to end through smoke(). Both new tests fail without the fix. The module gives 22 passed.
• Still to verify: the rest of the smoke (foreign() and the probe) has not run on real Windows 3.15. A build-only Conda Update run on this commit plus fix(release): read the environment without os.environb where Windows lacks it (#2120) #2122 and ci(conda): build the macos-26-intel legs again (#2123) #2124 will show it.

[JarryShaw]

GOOD TO GO at 0be0ba004 (Sonnet review).

• Upstream: CFEP-27's Windows lib/python/site-packages checks out against the python-feedstock recipe (python_site_packages_path). So does conda-build 26.9.1's get_site_packages, which reads that field into SP_DIR.
• Failure: the CI failure is exactly built_for returning python. foreign(), the platform check and the probe all run after it, so none of them ran.
• Layouts: POSIX lib/pythonX.Y[t], Lib/site-packages and lib/python/site-packages all classify correctly. A missing pyXY fails closed.
• Tests: both new tests fail with util/conda_vendor.py reverted, and the module gives 22 passed.
• Unverified: the rest of the smoke on real Windows 3.15. Build-only run 40043605163 on verify/conda-2126 (main + fix(release): read the environment without os.environb where Windows lacks it (#2120) #2122 + ci(conda): build the macos-26-intel legs again (#2123) #2124 + this) will show it.
• Conflict: this PR and fix(release): read the environment without os.environb where Windows lacks it (#2120) #2122 both edit the module docstring of tests/project/test_conda_windows_leg_2060.py. The resolution keeps both bullets, and I'll rebase whichever lands second.

[github-actions]

Coverage: 91.54% (unit tier, Python 3.15, 0be0ba004, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	4	20	2	95.16%
pcapkit/const	19599	880	2618	866	92.12%
pcapkit/corekit	2580	75	814	37	96.46%
pcapkit/dumpkit	282	3	102	3	98.44%
pcapkit/foundation	2944	23	1028	25	98.79%
pcapkit/interface	138	4	52	3	96.32%
pcapkit/protocols	17414	219	4574	185	98.14%
pcapkit/toolkit	869	36	298	10	95.20%
pcapkit/utilities	473	7	136	5	98.03%
pcapkit/vendor	4432	1988	1018	167	51.87%

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