The two adopters are the Windows measurement, and the probe list in
docs/cross-platform-port.md was written for someone fluent in a terminal.
This is the same seven questions in a form a private person can follow
top to bottom: one step per line, the exact command, the expected output
written out, and a four-part report form so a divergence arrives as the
command, the whole message and the Windows version rather than "it did
not work".
Every expected output here was measured rather than recalled: the
ls-remote hash, the guard's reported version, the bootstrap and launcher
refusal texts. The Windows-side claims that could not be measured from
this Mac -- setx persistence, the Microsoft Store execution alias, py.exe
absence on a Store install -- are cited from Microsoft's own docs, and
where the exact string was only attested secondhand the checklist asks
the tester to report what they actually saw.
Carries one blocker in an operator-only section: bootstrap.py installs
with --group <repo root>/pyproject.toml:dev, and pyproject.toml is not in
package_plugin.py's INCLUDE tuple, so the bootstrap cannot succeed in an
installed copy on any platform. The O4 gate then refuses to serve, which
would answer measurement 2 -- the critical one -- "no" for a reason that
has nothing to do with Windows. The fix is outside this order's fence;
the checklist names the pre-fix error text so a tester stops instead of
reporting a spoiled result.
Verify: .venv/bin/python -m pytest tests/ -> 223 passed, exit 0.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Kjell Tore Guttormsen
21d7f5a675