Skip to content
This repository was archived by the owner on Sep 9, 2026. It is now read-only.

closeout: fix #325 stale drive-test note; mark the REDESIGN program complete - #343

Merged
philoserf merged 2 commits into
mainfrom
wave-redesign-closeout
Jul 21, 2026
Merged

philoserf merged 2 commits into
mainfrom
wave-redesign-closeout

Conversation

@philoserf

Copy link
Copy Markdown
Owner

Closes the REDESIGN wave program: one live bug fix plus the doc/status closeout.

#325 — the stale pre-#300 note (bug)

TestDesignDriveStageCatalogP127's doc comment carried a leftover "Tracked in #300. Until it is settled, asserting either value here would dress a coin-flip as a golden" — residue from before the p.300 drive-cost ruling was settled. It contradicted its own comment block (which argues both catalogs must hold the Modified /2 cell) and the test itself (which asserts that cell). Left in place, it was the exact sentence a re-opener of the twice-mis-resolved ruling would cite as authority — in the file they'd be editing. The repo's signature fail-open defect, in its most expensive spot.

  • Deleted the two lines (drive_test.go). The protected /2 assertion is unchanged; both deliberately-redundant catalogs still pass.
  • Verified non-vacuous: the {Modified, …, "=7.5"} case still stands, grep dress → 0, TestDesignDriveStageCatalogP127/P134 both PASS.
  • Also qualified CLAUDE.md's unconditional "never re-introduce an mcr == 0-style sentinel" — P134 legitimately carries one for rows whose printed cost isn't reproducible under the final-tonnage rule, with the Modified cell corroborated in the separate exact-cost loop. The prohibition is against sentinelling away the /2 reading, not against that mechanism. (The issue's "worth a look while in there.")

REDESIGN.md — program complete (docs)

All seven structural findings resolved. Updated the Status table: §2 (#327) and §6 (#331) move partial → done (record codec + byte-identical sector save/load in PRs #337/#341; career.go split in PRs #339/#342). Both issues are being closed against their deferred remainders in other epics — bare references, no closing keyword: T5SS .sec import → #195, Species on Character → #185. The §4-general primitive tier (geom/#176, Money) stays unstarted by design.

Review

  • /simplify — N/A (a comment deletion + doc edits have no logic surface).
  • /code-review — skipped by judgment: the code change is a two-line comment deletion whose behavioral effect is verified above (protected assertion intact, tests green). Flagged here rather than silently.

Closes #325

https://claude.ai/code/session_01CnrrDqfE9MFUYxc1YZfGbz

…ssertion (#325)

TestDesignDriveStageCatalogP127's doc comment ended with a "Tracked in #300.
Until it is settled, asserting either value here would dress a coin-flip as a
golden" note — residue from before #300 was settled. It contradicted the body
of its own comment block (which explains why BOTH catalogs must hold the cell)
and the test itself (which asserts the Modified /2 cell at the case table). Left
in place it was the exact sentence a re-opener of the twice-mis-resolved p.300
ruling would cite as authority, sitting in the file they would edit.

Delete the two lines. The protected `/2` assertion is unchanged and both
redundant catalogs still pass. Also qualify CLAUDE.md's unconditional
"never re-introduce an mcr == 0-style sentinel" — P134 legitimately carries one
for non-reproducible rows, with the Modified cell corroborated in the separate
exact-cost loop; the prohibition is against sentinelling away the /2 reading.

Closes #325

Co-Authored-By: Claude
Claude-Session: https://claude.ai/code/session_01CnrrDqfE9MFUYxc1YZfGbz
All seven structural findings are resolved. Update the Status table: §2 (#327)
and §6 (#331) move from partial to done — the full Second Survey record codec +
byte-identical sector save/load landed (PRs #337/#341), and career.go was split
into five concern files (PRs #339/#342). Both issues are closed against their
deferred remainders in other epics (#195 .sec import, #185 Species). The
§4-general primitive tier stays unstarted by design — the next chapter, built
when a Maker forces it.

Co-Authored-By: Claude
Claude-Session: https://claude.ai/code/session_01CnrrDqfE9MFUYxc1YZfGbz
@philoserf
philoserf merged commit 7b867d7 into main Jul 21, 2026
1 check passed
@philoserf
philoserf deleted the wave-redesign-closeout branch July 21, 2026 03:42
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

drive_test.go carries pre-ruling residue that licenses deleting the #300 assertion it sits on

1 participant