Lens proposal: lift cross-concept nested cop helpers to top level under Cops::Sdlc::Structural

jared commented

2026-11-03 14:04:45 +00:00

Owner

Problem

Three helper objects are nested inside one cop but have callers in other cop concepts. Lens A2 says an object with callers in more than one concept is a top-level object. The nesting hides the shared role, and each new cop that needs a helper reaches into another cop's namespace. Structural::EdgeScope and Structural::EffectReach already show the correct top-level form.

Observed

The A2 lens challenge on #1511 (run 221) returned "code defect found". The contract-auditor had marked the break JUSTIFIED because the ticket asked for reuse of nested helpers; the challenge found that the ticket named SameFileDefs and EFFECT_CONSTANTS only, never Candidate, and that the other nested reuses are earlier A2 breaks, not a pattern to follow.

plugins/os-sdlc/lib/os_sdlc/cops/structural/domain_effect_in_edge/condition.rb:35,42  PrecedenceInEdge::Candidate.unwrap
plugins/os-sdlc/lib/os_sdlc/cops/structural/domain_effect_in_edge/facts.rb:15          QueryMethodReachesWrite::SameFileDefs
plugins/os-sdlc/lib/os_sdlc/cops/structural/precedence_in_edge.rb:17                   QueryMethodReachesWrite::SameFileDefs
plugins/os-sdlc/lib/os_sdlc/cops/structural/edge_scope.rb:14-15,65                     EffectInModel::EffectCall
plugins/os-sdlc/lib/os_sdlc/cops/structural/effect_reach/walk.rb:31                    EffectInModel::EffectCall

Reproduce

n/a: structure finding, the suite is the proof

Approach

Structure-only moves under RuboCop::Cop::Sdlc::Structural, one concept per commit, no behavior change, on the fast path.

Required work:

Move QueryMethodReachesWrite::SameFileDefs to structural/same_file_defs.rb and update its callers (QueryMethodReachesWrite, PrecedenceInEdge, DomainEffectInEdge::Facts). — check: present class SameFileDefs in plugins/os-sdlc/lib/os_sdlc/cops/structural/same_file_defs.rb
Move EffectInModel::EffectCall to structural/effect_call.rb and update its callers (EffectInModel, EdgeScope, EffectReach::Walk). — check: present class EffectCall in plugins/os-sdlc/lib/os_sdlc/cops/structural/effect_call.rb
Extract the class methods PrecedenceInEdge::Candidate.unwrap and .slot_of into a top-level value (name to be ruled) and update the callers (PrecedenceInEdge Candidate, Presence, Guard; DomainEffectInEdge::Condition). Candidate itself stays nested. — check: absent PrecedenceInEdge::Candidate in plugins/os-sdlc/lib/os_sdlc/cops/structural/domain_effect_in_edge/**/*.rb

Orchestrator work: rule the name of the unwrap value before item 3.

Acceptance criteria

Given the os-sdlc suite at the current commit, when the three moves land, then the suite and rubocop stay green with no test change.
Given any cop under cops/structural/, when it needs SameFileDefs, EffectCall, or the unwrap, then it references a top-level Structural:: constant, not another cop's nested constant.

Owner sketch

module RuboCop::Cop::Sdlc::Structural
  class SameFileDefs
  class EffectCall
  class UnwrappedNode  # name to be ruled
end

Origin

Trigger: /os-sdlc:implement 1511, run 221, contract-auditor LENS line on A2; challenged per references/lens-challenge-checklist.md
Improvised this session: none; signal line appended to plugins/os-sdlc/lens-signals/A2.jsonl (main ef991fc4)
Chain: DomainEffectInEdge::Condition calls PrecedenceInEdge::Candidate.unwrap ← the unwrap has no top-level home ← SameFileDefs and EffectCall were nested the same way when PrecedenceInEdge and EdgeScope reused them (DESIGN: lens-architecture A2; plugins/os-sdlc/lens-signals/A2.jsonl 2026-11-02 deferred row)
Root candidate: none found; searched SameFileDefs, EffectCall, Candidate.unwrap, nested helper, Lens proposal, top-level object, lift, Structural
Where: plugins/os-sdlc/lib/os_sdlc/cops/structural/
Session: a5c3e206-c44a-409e-8655-4a519fdaab84
Transcript: /home/jared/.claude/projects/-home-jared-dev-cc-os--claude-worktrees-ticket-1510/a5c3e206-c44a-409e-8655-4a519fdaab84.jsonl

VERDICT: CREATE — the A2 challenge on #1511 found a real code defect; no open ticket covers lifting the nested Structural cop helpers.

## Problem Three helper objects are nested inside one cop but have callers in other cop concepts. Lens A2 says an object with callers in more than one concept is a top-level object. The nesting hides the shared role, and each new cop that needs a helper reaches into another cop's namespace. `Structural::EdgeScope` and `Structural::EffectReach` already show the correct top-level form. ## Observed The A2 lens challenge on #1511 (run 221) returned "code defect found". The contract-auditor had marked the break JUSTIFIED because the ticket asked for reuse of nested helpers; the challenge found that the ticket named `SameFileDefs` and `EFFECT_CONSTANTS` only, never `Candidate`, and that the other nested reuses are earlier A2 breaks, not a pattern to follow. ``` plugins/os-sdlc/lib/os_sdlc/cops/structural/domain_effect_in_edge/condition.rb:35,42 PrecedenceInEdge::Candidate.unwrap plugins/os-sdlc/lib/os_sdlc/cops/structural/domain_effect_in_edge/facts.rb:15 QueryMethodReachesWrite::SameFileDefs plugins/os-sdlc/lib/os_sdlc/cops/structural/precedence_in_edge.rb:17 QueryMethodReachesWrite::SameFileDefs plugins/os-sdlc/lib/os_sdlc/cops/structural/edge_scope.rb:14-15,65 EffectInModel::EffectCall plugins/os-sdlc/lib/os_sdlc/cops/structural/effect_reach/walk.rb:31 EffectInModel::EffectCall ``` ## Reproduce n/a: structure finding, the suite is the proof ## Approach Structure-only moves under `RuboCop::Cop::Sdlc::Structural`, one concept per commit, no behavior change, on the fast path. Required work: 1. Move `QueryMethodReachesWrite::SameFileDefs` to `structural/same_file_defs.rb` and update its callers (QueryMethodReachesWrite, PrecedenceInEdge, DomainEffectInEdge::Facts). — check: present `class SameFileDefs` in `plugins/os-sdlc/lib/os_sdlc/cops/structural/same_file_defs.rb` 2. Move `EffectInModel::EffectCall` to `structural/effect_call.rb` and update its callers (EffectInModel, EdgeScope, EffectReach::Walk). — check: present `class EffectCall` in `plugins/os-sdlc/lib/os_sdlc/cops/structural/effect_call.rb` 3. Extract the class methods `PrecedenceInEdge::Candidate.unwrap` and `.slot_of` into a top-level value (name to be ruled) and update the callers (PrecedenceInEdge Candidate, Presence, Guard; DomainEffectInEdge::Condition). `Candidate` itself stays nested. — check: absent `PrecedenceInEdge::Candidate` in `plugins/os-sdlc/lib/os_sdlc/cops/structural/domain_effect_in_edge/**/*.rb` Orchestrator work: rule the name of the unwrap value before item 3. ## Acceptance criteria 1. Given the os-sdlc suite at the current commit, when the three moves land, then the suite and rubocop stay green with no test change. 2. Given any cop under `cops/structural/`, when it needs `SameFileDefs`, `EffectCall`, or the unwrap, then it references a top-level `Structural::` constant, not another cop's nested constant. ## Owner sketch ```ruby module RuboCop::Cop::Sdlc::Structural class SameFileDefs class EffectCall class UnwrappedNode # name to be ruled end ``` ## Origin - Trigger: `/os-sdlc:implement 1511`, run 221, contract-auditor LENS line on A2; challenged per references/lens-challenge-checklist.md - Improvised this session: none; signal line appended to plugins/os-sdlc/lens-signals/A2.jsonl (main ef991fc4) - Chain: DomainEffectInEdge::Condition calls PrecedenceInEdge::Candidate.unwrap ← the unwrap has no top-level home ← SameFileDefs and EffectCall were nested the same way when PrecedenceInEdge and EdgeScope reused them (DESIGN: lens-architecture A2; plugins/os-sdlc/lens-signals/A2.jsonl 2026-11-02 deferred row) - Root candidate: none found; searched SameFileDefs, EffectCall, Candidate.unwrap, nested helper, Lens proposal, top-level object, lift, Structural - Where: plugins/os-sdlc/lib/os_sdlc/cops/structural/ - Session: a5c3e206-c44a-409e-8655-4a519fdaab84 - Transcript: /home/jared/.claude/projects/-home-jared-dev-cc-os--claude-worktrees-ticket-1510/a5c3e206-c44a-409e-8655-4a519fdaab84.jsonl VERDICT: CREATE — the A2 challenge on #1511 found a real code defect; no open ticket covers lifting the nested Structural cop helpers.

jared added the

labels

2026-11-03 14:04:45 +00:00

jared referenced this issue

2026-11-03 14:04:55 +00:00

os-sdlc cop Sdlc/Structural/DomainEffectInEdge: an edge turns what it read into a verdict, a selection of work, or a next step (A19 R3) #1511

jared commented

2026-11-03 14:19:57 +00:00

Author

Owner

Ruling (Jared, 2026-11-03): item 3 name is Structural::Operand. Follow-up question under investigation before item 3: does DomainOperandInEdge (#1510) model the same operand concept, so that Operand should own more than unwrap/slot_of? Work starting on main via fast-batch.

Ruling (Jared, 2026-11-03): item 3 name is `Structural::Operand`. Follow-up question under investigation before item 3: does DomainOperandInEdge (#1510) model the same operand concept, so that Operand should own more than unwrap/slot_of? Work starting on main via fast-batch.

jared commented

2026-11-03 14:21:45 +00:00

Author

Owner

This is a design-only refactor with no observed failure, and its parent #1523 is itself an AI-created follow-up, so the depth-1 rule applies. If the Operand work in #1523 wants a third method for the test-call unwrap in Presence#subject_of and Condition#check_of, add it there. Note it on #1523 instead of filing a new ticket.