Fix #297: [Model] DisjointConnectingPaths

[GiggleLiu]

Summary

• add the implementation plan for the DisjointConnectingPaths model
• map the issue requirements onto the add-model workflow

Fixes #297

[codecov]

Codecov Report

❌ Patch coverage is 98.23789% with 4 lines in your changes missing coverage. Please review.
✅ Project coverage is 97.62%. Comparing base (6356007) to head (4870b41).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/models/graph/disjoint_connecting_paths.rs	97.36%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #746      +/-   ##
==========================================
+ Coverage   97.61%   97.62%   +0.01%     
==========================================
  Files         435      437       +2     
  Lines       53770    53997     +227     
==========================================
+ Hits        52485    52715     +230     
+ Misses       1285     1282       -3     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[GiggleLiu]

Implementation Summary

Changes

• Added the new DisjointConnectingPaths<SimpleGraph> model, including deterministic edge-order encoding, constructor validation, canonical example-db fixture, and focused unit tests.
• Wired the model into the graph/model exports and prelude so it is discoverable through the registry and example exporters.
• Added CLI creation support via pred create DisjointConnectingPaths --graph ... --terminal-pairs ..., including parser validation and targeted CLI tests.
• Documented the model in docs/paper/reductions.typ and added the Robertson--Seymour / Kawarabayashi--Kobayashi--Reed bibliography entries used by the paper entry.

Deviations from Plan

• The plan mentioned a manual alias update in problem_name.rs, but that was unnecessary: catalog alias lookup is already case-insensitive over canonical names, so DisjointConnectingPaths is discoverable without extra hand-written mapping.

Open Questions

• None.

[GiggleLiu]

Agentic Review Report

Structural Check

Structural Review: model DisjointConnectingPaths

Structural Completeness

#	Check	Status
1	Deterministic whitelist	FAIL — problemreductions-cli/src/cli.rs, problemreductions-cli/src/commands/create.rs, and src/lib.rs are outside the expected whitelist for a model PR
2	Deterministic completeness	PASS
3	Model file exists	PASS — src/models/graph/disjoint_connecting_paths.rs
4	inventory::submit! present	PASS
5	Serialize/Deserialize derive present	PASS
6	Problem trait impl present	PASS
7	SatisfactionProblem impl present	PASS
8	Test link present	PASS — #[cfg(test)] + #[path = ...] present
9	Test file exists	PASS — src/unit_tests/models/graph/disjoint_connecting_paths.rs
10	Test file has >= 3 test functions	PASS — 10 test functions
11	Registered in src/models/graph/mod.rs	PASS
12	Re-exported in src/models/mod.rs	PASS
13	declare_variants! entry exists	PASS
14	CLI alias resolution	PASS — generic catalog-based resolve_alias() covers canonical names without a hand-written alias arm
15	CLI create support	PASS
16	Canonical model example registered	PASS — added via graph::canonical_model_example_specs()
17	Paper display-name entry	PASS
18	Paper problem-def block	PASS
19	Blacklisted autogenerated files absent	PASS

Build Status

• make test: PASS
• make clippy: PASS

Semantic Review

• evaluate() correctness: OK — selected edges must decompose into exactly k path components whose endpoints match the requested terminal pairs; cycles, branching, reused terminals, wrong-length configs, and non-binary entries are rejected.
• dims() correctness: OK — vec![2; num_edges()] matches the edge-subset encoding described in the issue.
• Size getter consistency: OK — num_vertices(), num_edges(), and num_pairs() align with the schema and complexity metadata.
• Weight handling: OK / N/A — this is an unweighted satisfaction model.

Issue Compliance

#	Check	Status
1	Problem name matches issue	OK
2	Mathematical definition matches	OK
3	Problem type (sat) matches	OK
4	Type parameters match	OK — SimpleGraph
5	Configuration space matches	OK — one binary variable per edge
6	Feasibility check matches	OK
7	Objective function matches	OK — boolean satisfiability
8	Complexity matches	ISSUE — the code uses the requested 2^num_edges complexity string, but the paper text cites @karp1972 for NP-completeness instead of the linked issue’s approved ND40 / Knuth-Karp-Lynch trail

Summary

• 18/19 structural checks passed
• 7/8 issue-compliance checks passed
• FAIL — deterministic whitelist violation: problemreductions-cli/src/cli.rs, problemreductions-cli/src/commands/create.rs, and src/lib.rs are outside the expected model-PR whitelist.
• ISSUE — docs/paper/reductions.typ:973-975 documents pred solve disjoint-connecting-paths.json, but the model has no reduction path to ILP, so the documented command fails unless --solver brute-force is added.
• ISSUE — docs/paper/reductions.typ:969 does not match the linked issue’s approved NP-completeness references.

---

Quality Check

Quality Review

Design Principles

• DRY: OK — the implementation follows existing model/create-command patterns without obvious copy-paste debt.
• KISS: OK — the model uses a single edge-subset validator with straightforward graph checks.
• HC/LC: OK — graph semantics stay in the model, while CLI parsing stays in create.rs.

HCI (CLI changed)

• Error messages: OK — invalid terminal-pair inputs return specific messages plus usage guidance.
• Discoverability: OK — pred list, pred show, and pred create --example all surface the new model cleanly.
• Consistency: OK — --terminal-pairs fits the existing graph-model CLI patterns.
• Least surprise: ISSUE — the paper’s documented command pred solve disjoint-connecting-paths.json fails by default because DisjointConnectingPaths has no ILP path; users must infer --solver brute-force from the error instead of from the docs.
• Feedback: OK — the CLI reports both the missing path and the brute-force workaround clearly.

Test Quality

• Naive test detection: ISSUE
  • The added tests cover model validation and CLI JSON creation, but nothing exercises the documented end-user solve flow on the canonical example. That gap let the broken paper command ship without a guardrail.

Issues

Critical (Must Fix)

• None.

Important (Should Fix)

• docs/paper/reductions.typ:973-975 documents a default pred solve command that fails on the canonical example. Confirmed locally: pred path DisjointConnectingPaths ILP returns No reduction path from DisjointConnectingPaths to ILP, and pred solve <example> exits with Try --solver brute-force.

Minor (Nice to Have)

• docs/paper/reductions.typ:969 uses an NP-completeness citation that does not match the linked issue’s approved references.
• No automated test currently protects the documented solve flow for the new model.

Summary

• ISSUE — the paper’s solve snippet is not reproducible as written.
• ISSUE — tests do not cover the documented solve flow that regressed.
• ISSUE — the paper’s hardness citation diverges from the linked issue review.

---

Agentic Feature Tests

Feature Test Report: problem-reductions

Date: 2026-03-22
Project type: CLI tool / Rust workspace
Features tested: DisjointConnectingPaths model discovery, inspection, creation, solving, and evaluation
Profile: ephemeral read-only review run
Use Case: A downstream CLI user tries to discover the new model from the catalog, create the canonical example, solve it, and confirm the documented commands work as written.
Expected Outcome: pred list, pred show, pred create --example, pred create, pred solve, and pred evaluate should work end-to-end for DisjointConnectingPaths.
Verdict: fail
Critical Issues: 1

Summary

Feature	Discoverable	Setup	Works	Expected Outcome Met	Doc Quality
DisjointConnectingPaths CLI flow	yes	yes	partial	no	broken solve snippet

Per-Feature Details

DisjointConnectingPaths CLI flow

• Role: downstream CLI user validating a newly added model
• Use Case: discover the model, inspect its metadata, create the canonical example, solve it, and verify the witness config
• What they tried: pred list, pred show DisjointConnectingPaths, pred show disjointconnectingpaths, pred create --example DisjointConnectingPaths, pred create DisjointConnectingPaths --graph ... --terminal-pairs ..., pred path DisjointConnectingPaths ILP, pred solve <example>, pred solve <example> --solver brute-force, and pred evaluate <example> --config 1,0,1,0,1,0,1
• Discoverability: Good. The model appears in pred list, pred show works, and lowercase canonical lookup also works.
• Setup: Good. make cli succeeded and the installed pred binary ran without extra setup.
• Functionality: Partial. Listing, showing, creating, brute-force solving, and evaluating all worked. Default solving failed because there is no reduction path from DisjointConnectingPaths to ILP.
• Expected vs Actual Outcome: Expected the documented paper command pred solve disjoint-connecting-paths.json to work. Actual result: exit status 1 with Error: No reduction path from DisjointConnectingPaths to ILP or ILP solver found no solution. Try --solver brute-force.
• Blocked steps: None.
• Friction points: The paper suggests a command that does not work unless the user adds --solver brute-force.
• Doc suggestions: Change the paper snippet to pred solve disjoint-connecting-paths.json --solver brute-force, or explicitly explain that the model currently has no ILP path.

Expected vs Actual Outcome

The catalog and creation flows matched expectations. The documented solve flow did not: the canonical example only solves with --solver brute-force, so the current paper snippet is not reproducible.

Issues Found

• Confirmed, important: the documented default solve command fails because DisjointConnectingPaths has no ILP reduction path.

Suggestions

• Update the paper command block to use --solver brute-force.
• Add a CLI/documentation smoke test that exercises the canonical example with the documented commands.

---

Generated by review-pipeline