State the handoff reader contract nomologR asked for, and record the joint 1.0 checklist

[JUhalt]

nomologR asked for three handoff clarifications before schema 1 becomes a 1.0 promise, and proposed a release-candidate checklist for the joint 1.0. The maintainer approved all of it on 2026-09-27. The schema and every handoff are unchanged, and no computed value changes.

1. "Reading the decisions", a new section in ?content_handoff

A reader takes each item's decision from carried and status, and never re-derives it by comparing value with criterion or by branching on provenance$package_version. status is one of the four values keep accepts; recommendation words the same decision for a person and, like rule, is prose.

The example is the nine-expert correction in 0.8.0. An item that 7 of 9 experts rated relevant has an I-CVI of .778 in both releases:

• 0.7.0 compared it with a rounded .78 and held it back;
• 0.8.0 applies Lynn's (1986) 7 of 9, stores 7/9 as the criterion, and carries it.

In the 0.8.0 handoff the value equals the criterion, so a recomputed value >= criterion would hang on floating-point rounding, where the package compares counts of experts. The section also restates that keying of NA means unknown, never forward-worded. The page now lists Lynn (1986) under References.

2. An item sits in at most one scale

The scales entry now says so: an item belongs to at most one scale, the one its item_evidence$scale names, because every workflow that maps items to constructs requires exactly one target per item. It also notes that a Delphi handoff has scales = NULL.

New tests:

• for the sort, rating, and congruence workflows, no item appears in two scales, and each carried item sits in the scale its evidence row names;
• each of the three workflows refuses an item given two targets;
• every workflow's status (including Delphi and essentiality) is one of the values keep accepts, and carried follows status and keep exactly.

nomologR's reader keeps its own check for an item in two scales, since a hand-built or edited object can still contain one.

3. A "checked, none reversed" fixture (delivered outside the repo)

nomologR's reader treats keying = 1 throughout ("someone checked, none reversed") differently from NA ("nobody said"), and tested that case by editing a 0.7.0 object. The fixture generator gained a fit with reverse_keyed = character(0) and response_scale = c(1, 5), run from the v0.9.0 tag. The new file equals the delivered walkthrough-sort file in every field except keying and created. The fifteen files nomologR already had are byte-identical. nomologR added it in JUhalt/nomologR#102.

4. The joint 1.0 checklist, in the ROADMAP

The v1.0.0 section records the release-candidate checklist, in addition to nomologR#53's criteria:

• a final fixture set from each release candidate, with md5 sums, that nomologR's tests pass before either package tags;
• the compatibility table in both READMEs (contentvalidR 1.x writes schema 1, nomologR 1.x reads it), added at the release candidate so neither README describes a 1.x before one exists;
• one handoff example in both READMEs, regenerated from contentvalidR's release candidate;
• the joint walkthrough in one package, with the other linking to it. Not yet decided. Recommended, and acceptable to nomologR: vignette("one-item-set-both-stages") here, linking to nomologR's guided-workflow article. The maintainer decides;
• release notes that link each other.

The CRAN criterion now notes that nomologR 0.3.0 was submitted on 2026-09-26.

Verification

• All 2,975 tests pass (2,950 before, plus the new expectations). Spelling is clean.
• The rendered help text was read for the new section.
• tools/release-gate.R on this PR's head (33b106f), run in a detached worktree: build, smoke, and check all pass. R CMD check: 0 errors, 0 warnings, and only the two expected notes (New submission; math rendering skipped without V8).

🤖 Generated with Claude Code