Twinned from https://github.com/cfsh/ninefold/issues/608 by tools/beads/import_github.py's reconcile pass.
Suggestions from ninefold-reviewer's review of #603. Advisory only — none of these blocked that PR, and the measure-the-box change it makes is sound.
1. "The full gate runs here" is asserted on a box with no shellcheck
tools/office/spawn_dev.py, Box.gate_note. can_merge is now dotnet and godot, so a box with the SDK and the engine but no shellcheck is told: *"The full gate runs here: ./presubmit.py --all runs python, dotnet, and Godot is installed, so probes run and YOU CAN MERGE."*
notes/machines.md documents that configuration as real — shellcheck is per-user on the WSL box (~/.local/bin), so a second account there has dotnet and Godot machine-wide and no shell lane. can_merge is correct in that case (--all still passes; the shell lane degrades to its loud skip), so this is wording rather than a wrong predicate. But "the full gate" is the #389 shape the Box docstring itself warns against — *"a gate that prints green while doing less than the reader assumes"* — and the sentence's own lane list quietly contradicts it. Either gate the phrase on len(self.suites) == 3, or name the absent lane. No test covers dotnet + Godot + no-shellcheck; the new FULL_BOX has all three.
2. The last role literal, in the sibling test file
tools/tests/test_spawn_dev.py:294 still reads *"the WINDOWS sandbox box — the only box that can run --all, and therefore the only box that can merge (#558 F1)"*. That is the same present-tense claim #603 corrected at spawn_dev.py:187, in the file it also edits. (spawn_dev.py:984 — *"red on the Windows merge box"* — reads as the history of one specific past failure, so it is arguably fine as written.)
3. Godot is detected only where nflib.godot.find() globs
read_box asks nflib.godot.find(), which is the right call — no fourth spelling of the path (#571). The consequence worth recording: a box whose engine is reachable only through GODOT_EXE / GODOT_BIN (overrides probe.py and f5.py honour) is told it cannot merge when it can. It under-claims rather than over-claims, which is the safe direction, and the new comment states the choice deliberately. Filed so #571 remembers this call site when it unifies the spellings.
4. After #603 closes #601, the machines.md staleness is tracked only inside a PR
Issue #601 also asked for notes/machines.md, "which now has to describe more than two machines". #603 correctly defers that to #549 — but #549 is an open PR with no backing issue, and #601 closes on merge. If #549 is reshaped or closed, nothing tracks that ## The environments still says "uname -s = Linux → the WSL office box" while a non-WSL Linux box is in the studio. Worth a line on #549, or a small unclaimed issue.
None.
No comments.